Institutionen för systemteknik

Size: px
Start display at page:

Download "Institutionen för systemteknik"

Transcription

1 Institutionen för systemteknik Department of Electrical Engineering Examensarbete Grey-box Modelling of a Quadrotor Using Closed-loop Data Examensarbete utfört i reglerteknik vid Tekniska högskolan vid Linköpings universitet av Marcus Bäck LiTH-ISY-EX--15/4911--SE Linköping 2015 Department of Electrical Engineering Linköpings universitet SE Linköping, Sweden Linköpings tekniska högskola Linköpings universitet Linköping

2

3 Grey-box Modelling of a Quadrotor Using Closed-loop Data Examensarbete utfört i reglerteknik vid Tekniska högskolan vid Linköpings universitet av Marcus Bäck LiTH-ISY-EX--15/4911--SE Handledare: Examinator: Daniel Simon isy, Linköpings universitet Ola Härkegård Saab Aeronautics Peter Rosander Saab Aeronautics Martin Enqvist isy, Linköpings universitet Linköping, 27 november 2015

4

5 Avdelning, Institution Division, Department Datum Date Department of Automatic Control Department of Electrical Engineering SE Linköping Språk Language Rapporttyp Report category ISBN Svenska/Swedish Licentiatavhandling ISRN Engelska/English Examensarbete C-uppsats D-uppsats LiTH-ISY-EX--15/4911--SE Serietitel och serienummer Title of series, numbering Övrig rapport ISSN URL för elektronisk version Titel Title Författare Author Grey-box Modelling of a Quadrotor Using Closed-loop Data Marcus Bäck Sammanfattning Abstract In this thesis a quadrotor is studied and a linear model is derived using grey-box estimation, a discipline in system identification where a model structure based on physical relations is used and the parameters are estimated using input-output measurements. From imu measurements and measured pwm signals to the four motors, a direct approach using the prediction-error method is applied. To investigate the impact of the unknown controller the two-stage method, a closed-loop approach in system identification, is applied as well. The direct approach was enough for estimating the model parameters. The resulting model manages to simulate the major dynamics for the vertical acceleration and the angular rates well enough for future control design. Nyckelord Keywords System identification, Grey-box model, closed-loop identification, Two-stage method, multirotor, quadrotor, VTOL, IMU

6

7 Abstract In this thesis a quadrotor is studied and a linear model is derived using grey-box estimation, a discipline in system identification where a model structure based on physical relations is used and the parameters are estimated using input-output measurements. From imu measurements and measured pwm signals to the four motors, a direct approach using the prediction-error method is applied. To investigate the impact of the unknown controller the two-stage method, a closed-loop approach in system identification, is applied as well. The direct approach was enough for estimating the model parameters. The resulting model manages to simulate the major dynamics for the vertical acceleration and the angular rates well enough for future control design. iii

8

9 Acknowledgments Without the extensive support from family, friends and colleagues I would not have managed to complete my studies. A special thank you is directed to my supervisors at Saab and isy, who have found the time to assist me when needed, and actively asked how my work progressed. My opponent has provided me with help and support at moments notice. Finally, I would like to thank my examiner. His interest in my work, and enthusiasm when teaching, has been a huge motivation and inspiration for a soon-to-be engineer. This thesis is dedicated to all who have helped me, and the king of Sweden. Live long and prosper. Linköping, November 2015 Marcus Bäck v

10

11 Contents Notation ix 1 Introduction Background Problem Formulation Limitations Related Work Outline System Identification Model Structures Identification Methods Closed-loop Identification Model Validation Experiment Design The Quadrotor System Motion of a Quadrotor Measured Signals Platform Development Physical Model States Kinetics Force Equations Torque Equations Motor and Rotor Dynamics Complete Motion Model Linearization Model Separation System Identification of the Vertical Dynamics Black-box Identification vii

12 viii Contents 5.2 Grey-box Identification Model Comparison System Identification of the Rotational Dynamics Roll Rate Identification Black-box Identification Direct Identification Two-stage Identification Model Comparison Pitch Rate Identification Yaw Rate Identification Final Model 49 8 Conclusions Future Development A Thrust Linearization 57 B Platform Development 59 Bibliography 63

13 Notation Symbols Symbol Description Θ Unknown parameter vector E The earth-fixed coordinate frame B The body-fixed coordinate frame X The quadrotor state vector U The input signal vector φ Roll angle θ Pitch angle ψ Yaw angle p Roll rate q Pitch rate r Yaw rate ξ E Position vector (x, y, z) T in the earth-fixed frame η E Angular position (φ, θ, ψ) T in the earth-fixed frame ν B Linear velocity (u, v, w) T in the body-fixed frame ω B Angular velocity (p, q, r) T in the body-fixed frame u i Input signal to motor i {1, 2, 3, 4} u T Input signal for collective thrust command u φ Input signal for roll command u θ Input signal for pitch command u ψ Input signal for yaw command T i Thrust force from motor i = {1, 2, 3, 4} T Collective thrust force T φ Thrust for producing a roll movement T θ Thrust for producing a pitch movement T ψ Thrust for producing a yaw movement M x Final model. x {w, φ, θ, ψ} ix

14 x Notation Abbreviations Abbreviation aic arx imu oe pem pwm std vtol Description Akaike s information criterion Auto-regressive with exogenous variable Inertial measurement unit Output error Prediction-error method Pulse width modulation Standard deviation Vertical take-off and landing

15 1 Introduction In this master s thesis, a mathematical model for the dynamics of a quadrotor platform is derived in collaboration with Saab Aeronautics. This is done using a semi-physical, or grey-box, approach where a physical model is derived and used as model structure for system identification where measured flight data is used to estimate the parameters of the model. The model is estimated using the System Identification Toolbox in Matlab [Ljung, 2014]. 1.1 Background Multirotor platforms, such as quadrotors, are today common in various applications. The simple mechanics of a quadrotor makes the platform agile and easy to control. Its ability for vertical take off and landing (vtol) is desirable in many applications such as flight in tight terrain. One application for a quadrotor is surveillance or reconnaissance in hazardous or otherwise inaccessible locations. Multirotors are also commonly used by media and offer possibilities for video coverage of large sports events or accidents. Due to the simplicity of the mechanics, the quadrotor (and other multirotor platforms) is a common hobby product for radio controlled flight pilots. This thesis is a continuation on a previous thesis where a quadrotor platform was constructed and an avionics solution for rapid prototyping and data logging was implemented [Blomberg, 2015]. The current quadrotor is provided with a commercial flight control system and this thesis is the first stepping stone towards designing a new flight controller. When designing a controller for an application, a model is needed. The model describes which behaviour is expected from the system and how different inputs affect the outputs. The purpose of this thesis is to derive a model describing the dynamics to be used in future control design. 1

16 2 1 Introduction 1.2 Problem Formulation The goal of the thesis is to derive a model for the dynamics of a hovering quadrotor using a grey-box approach to system identification. The rotors of a multirotor are usually at a fixed pitch, which is the case of the quadrotor used in this thesis. Because of the fixed rotors a model describing the body-fixed velocity along the rotor axis is desirable from a control design point of view. A common requirement when designing controllers for multirotor platforms is orientation, or attitude, control. Thus, the main focus of this thesis is to estimate the models for roll, pitch and yaw maneuvers together with a model for the vertical velocity. Due to its symmetry, the roll and pitch dynamics of a quadrotor should be similar. If this is the case, the same controller could be used for roll and pitch control. This is desirable due to the simplicity of the controller. In this thesis, the following problems will be answered: Can a grey-box approach be used to achieve a dynamic model for the quadrotor? Which model structure is suitable when modelling the dynamics of a hovering quadrotor? Can the motor dynamics be estimated using just the input signals to the motors and the measured accelerations and angular rates? Is it possible to use the same model for the roll and pitch dynamics? Is it possible to disregard the current controller and use a direct approach when identifying the system? 1.3 Limitations As stated above, the model to be derived will be limited to the vertical velocity as well as the three angular rates. Furthermore, the model will be linearized for a hovering quadrotor. A hovering quadrotor is floating mid-air at an arbitrary position. The model may therefore not be valid for a flight case too far from its operating region. Regarding the system limitations, all commands sent to the quadrotor will be generated by hand by a human pilot. No arbitrary reference signal may be chosen, other than what the pilot manages to produce. For example, when a chirp is wanted as input for the roll rate, it will not be possible to produce an ideal chirp, but the command will be as chirp-like as the pilot manages to achieve. 1.4 Related Work Modelling of vtol platforms is frequent in literature. Mettler [2003] describes an approach to developing a simple and effective linear parameterized model of small rotorcraft dynamics. It contains information on experiment design and

17 1.5 Outline 3 data collection, frequency response system identification and grey-box modelling, among other things. Most literature presents a white-box approach when modelling the multirotor. In Mahony et al. [2012] a physical model of a quadrotor is derived. It also presents state estimation using an inertial measurement unit (imu). Bouabdallah [2007] provides an extensive white-box approach including rigid body mechanics as well as multiple aerodynamic effects. Pounds, Mahony and Corke develop a quadrotor robot described in multiple papers, starting with a physical model of the rigid body mechanics [Pounds et al., 2004]. They continue with system identification of the motor dynamics [Pounds et al., 2007] and aerodynamic effects [Pounds et al., 2006]. A direct closed-loop black-box approach to estimate the dynamics of a hexarotor platform is taken in Magnusson [2014]. Magnusson also compares the blackbox model with a white-box model derived in the same master s thesis. In Lundmark and Lindblom [2015], a grey-box approach to estimate the dynamics of a hexarotor platform is presented. Lundmark and Lindblom present a physical model of the rigid body mechanics while a grey-box approach is taken for the aerodynamics and motor dynamics. Bergman and Ekström [2014] present a similar approach as Lundmark and Lindblom but applied to an octacopter. In Li [2014], a grey-box approach to estimate the dynamics of a quadrotor is presented. Li uses extensive sensor networks to estimate a non-linear model. For the non-linear part of the model, both an extended and an unscented Kalman filter is used. For the linear parts, the prediction-error method applied and a Box-Jenkins structure is used. Grey-box modelling of a miniature helicopter using the prediction-error method is presented in Zhou and Ping [2008]. Khorani et al. [2013] provide an online model estimator of a quadrotor platform based on measurements from an imu. The methods for system identification used in this thesis are based on the theories in Ljung [1999]. 1.5 Outline This thesis is divided into eight chapters. In Chapter 2 the theories of system identification used here are presented. In the following chapter, an overview of the quadrotor system and a short description of the hardware is given. This chapter also includes a brief explanation of the concepts of quadrotor maneuvering. At the end of Chapter 3 some results of the platform development during this thesis are presented. A physical model of the quadrotor is derived and linearized in Chapter 4. The model is used as grey-box structure in the identification chapters that follow. These chapters are divided into estimation of the vertical velocity model in Chapter 5 and estimation of the angular rates dynamics in Chapter 6. The two identification chapters result in four models, describing the vertical velocity, the roll rate, the pitch rate and the yaw rate, respectively. These models are summarized and combined into a final model in Chapter 7. In Chapter 8

18 4 1 Introduction some final conclusions are presented and some future development is suggested.

19 2 System Identification In many applications, a mathematical model of a dynamic system is desired. One way is to make a physical model, also known as white-box model, where the laws of physics are used to form the dynamic relations of a system. A significant problem with white-box modelling is that the equations quickly get very complex since many relations need to be accounted for. To minimize the complexity of the model some behaviour might be neglected or approximated in some way, resulting in a less accurate model. Another way to acquire a dynamic model describing a system is called system identification. A common approach is called black-box modelling. For a blackbox model, input-output measurements are collected and a dynamic model is estimated using these measurements with no concern taken to the underlying physics. The black-box modelling approach is popular due to its simplicity and that the same approach and methods can be used in different applications. A drawback is that the resulting model does not have the intuitive model structure of a white-box model, but consists of a set of equations that describes the inputoutput relations. A combination of the white- and black-box modelling gives the grey-box modelling approach, which requires some physical insight, but the values of unknown parameters can be estimated. If there is no physical knowledge of a part of the system, a black-box approach could be taken for that specific part. The resulting model will have the more intuitive structure of a white-box model but some of the flexibility of a black-box model. Also, since the parameters are estimated, the model takes into consideration some simplifications made when the physical structure was derived. A drawback in comparison to the black-box approach is that the model is limited to the chosen structure. If a bad structure is chosen, the model might be unable to describe the full dynamics of the system where instead a pure black-box might have captured the interesting dynamics. On the other 5

20 6 2 System Identification u G y r + Σ - F u G y (a) Open-loop system. A system without output feedback. (b) Closed-loop system where F is the controller and r is the reference signal. The output, y, is fed back to the controller. Figure 2.1: Examples of an open- and a closed-loop system where u is the input signal and y is the output signal to the system G. hand, if the gathered data contains lots of disturbances, a grey-box model might be able to limit the model to the true dynamics, while a black-box model might include some properties of the disturbances in the model instead. 2.1 Model Structures A system is a vague definition meaning anything that we wish to observe, in this thesis a quadrotor. The system in Figure 2.1a is an example on an open-loop system, i.e., a system were the input, u, influences the output, y, but not the opposite. In many cases, the system is not stable and in need of a controller to assure stability, and the quadrotor is one example of this. Without a controller the quadrotor is difficult (nearly impossible) to operate. A system with feedback from the output to a controller is called a closed-loop system and a simple example is shown in Figure 2.1b. The dynamics of a system is often described as a set of (linear or non-linear) differential equations ẋ = f (x, u) + v (2.1) where x is the state vector of the system, u is the input and v is process noise (such as wind). It is often not possible to measure the states exactly (or even directly) and the measurements are gathered in the equation y = h(x, u) + e (2.2) where y is the measured output, h( ) is a linear or non-linear function of x and u and e is additive measurement noise. These equations combined describe the dynamics of the entire system. This structure is referred to as a state-space model and for linear systems the state-space model can be written as ẋ = Ax + Bu + v y = Cx + Du + e (2.3) where A, B, C, D are matrices describing how each state or input signal affects the system. Here, ẋ is the time derivative of the state vector, x. For an unknown system these matrices contains unknown parameters and when a model

21 2.1 Model Structures 7 is wanted, the task is to estimate these parameters. In literature, the unknown parameters are often collected in a parameter vector denoted θ, but since θ is being used for pitch angle in this master thesis the parameter vector will be denoted Θ to avoid confusion. A system can be described as a continuous-time system or as a discrete-time system. The state-space model in (2.3) is presented as a continuous-time system. In some applications, a discrete-time model is a more convenient representation, for example if the model represents a sampled system. The simplest description of a discrete-time model structure is as a linear difference equation y(t) + a 1 y(t 1) + + a na y(t n a ) = b 1 u(t 1) + + b nb u(t n b ) + e(t) (2.4) which describes how previous inputs, u(t 1)... u(t n b ), and previous outputs, y(t 1)... y(t n a ) affect the output, y(t). For this model, the parameter vector is Θ = ( ) T a 1,..., a na, b 1,..., b nb. If we also define the regression vector ϕ(t) = ( y(t 1),..., y(t n a ), u(t 1),..., u(t n b ) ) T we can express the output as y(t) = Θ T ϕ(t) + e(t) = ϕ(t) T Θ + e(t) (2.5) which is called a linear regression. From (2.4) we introduce A(q) = 1 + a 1 q a na q n a and and form B(q) = b 1 q b nb q n b G(q, Θ) = B(q) 1, H(q, Θ) = A(q) A(q) which is the input-output transfer function and the noise-output transfer function, respectively. Here, q is the shift operator. This particular model structure is called an auto-regressive with exogenous input (arx) structure and it is visualized as a block diagram in Figure 2.2a. By expressing the transfer functions G(q, Θ) and H(q, Θ) differently, other common model structures can be formed. For example, the Box-Jenkins model structure has G(q, Θ) = B(q) F(q) and H(q, Θ) = C(q). A special case of the Box-Jenkins model structure is the output D(q) error (oe) model structure, where C(q) = D(q) = 1. The oe model structure is shown in Figure 2.2b.

22 8 2 System Identification e e u B Σ 1 A y u B F w Σ y (a) arx structure (b) oe structure Figure 2.2: Two different linear model structures. 2.2 Identification Methods When a model structure is chosen, the parameters are to be estimated. A common method for estimating the values of the parameters is the prediction-error method (pem) [Ljung, 1999, p ]. The model presents us with a prediction for the next value of y(t). For the model in (2.5), the predicted value is defined as ŷ(t Θ) = ϕ T Θ. The estimate ŷ(t Θ), can be compared with the next measurement, y(t), and by creating the prediction error ε(t, Θ) = y(t) ŷ(t Θ) (2.6) we can find the values for the parameter vector that most accurately describes the true system with respect to the prediction error. Assuming that we have sampled the input and output signals N times, we can then form the loss function V N (Θ) = 1 N N l (ε (t, Θ)) (2.7) t=1 where l( ) is a scalar function. A common choice of function is the quadratic function l(ε(t, Θ)) = ε(t, Θ) T ε(t, Θ) (2.8) resulting in the least squares method. A parameter estimate, ˆΘ, can then be found by minimizing the loss function, V N, over Θ as ˆΘ N = arg min V N (Θ). (2.9) Θ For a linear regression model, for example (2.5), the loss function is given by V N (Θ) = 1 N N ( y Θ T ϕ ) T ( y Θ T ϕ ) = 1 N t=1 N y T y 2Θ T f N + Θ T R N Θ (2.10) t=1

23 2.3 Closed-loop Identification 9 where and if R N is invertible f N = 1 N R N = 1 N N ϕy (2.11) t=1 N ϕϕ T (2.12) t=1 V N (Θ) = 1 N N t=1 y T y fn T R 1 N f N + ( Θ R 1 N f ) T ( N RN Θ R 1 N f ) N. (2.13) Since R N is a positive definite matrix in this case, the last term in the expression above is always positive. The loss function is therefore minimized when and thus ( Θ R 1 N f ) T ( N RN Θ R 1 N f ) N = 0 ˆΘ = R 1 N f N. The solution to the pem is normally not found analytically as above. Instead an iterative solution on the form ˆΘ (i+1) = ˆΘ i µr 1 N V N ( ˆΘ i ) (2.14) is applied. Here, ˆΘi denotes the i-th iteration estimate of the parameter vector Θ. Further, µ is the step size of the iteration, R N is (the approximation of) the Hessian of the loss function and V N (Θ) is the gradient of the loss function with respect to the parameter vector [Gustafsson et al., 2010, p ]. Each iteration results in a better estimate of the parameter vector Θ. The solution is acquired when a certain stop criterion is reached. A stop criterion could, for example, be when the difference of the parameter vector between two samples is small. For a state-space model, such as (2.3), the optimal parameters using the pem are found using an observer together with an iterative solution [Gustafsson et al., 2010, p. 243]. Another possible approach is to first transform the state-space model to a set of transfer functions and solve as above [Ljung, 2002]. 2.3 Closed-loop Identification For identification of a closed-loop system, there are some possible difficulties since the input to the system, u, is affected by the output, y. One is the risk of a too simple controller, which may result in non-informative experiments even if the input is persistently excited [Ljung, 1999, p. 430]. However, as long as the reference signal is persistently excited the closed-loop experiment is informative [Ljung, 1999, p. 432]. System identification for a closed-loop system is mentioned in Section 13.5 in Ljung [1999] and it is suggested that treating the system

24 10 2 System Identification direct as an open-loop system, should be the first approach. The same predictionerror methods used for open-loop system identification are applied to data from the closed-loop system. In general, a reqirement is that the model needs to be provided with a sufficiently rich noise model. However, in cases where the signal to noise ratio is high, the direct approach should be able to estimate the model well also when it contains only a simplified noise model. The direct approach is the most straight-forward approach in closed-loop identification since it uses the input, u, and the output, y, but takes no consideration to the reference signal, r, causing the approach to be indifferent to the complexity of the controller. If the direct approach fails, consideration to the closed loop needs to be taken. If the controller is unknown, a joint input-output identification approach is one option [Ljung, 1999, p. 438]. One particular method is the two-stage method, described in Van Den Hof et al. [1992]. Consider the closed-loop system in Figure 2.1b and assume it to be a single-input single-output system for simplicity. The input to the system can be expressed as u(t) = F r(t) + w(t) = Sr(t) + w(t) (2.15) 1 + FG where w is the combined effect of disturbances on the control signal. Using the measured input signal, u(t), and the known reference signal, r(t), a model for S can be estimated. From the model, Ŝ, the input signal can be simulated as û(t) = Ŝr(t) (2.16) and the simulated input can in turn be used to estimate the model of the system as y(t) = G(q, Θ)û(t) + e(t) (2.17) where e(t) is noise correlated with w(t). The sequence ū(t) = u(t) û(t) is uncorrelated with the sequence r(t), assuming that Ŝ is of high enough order. Suppose that the true system is given by y(t) = G 0 (q)u(t) + n(t) (2.18) where n(t) is noise that is uncorrelated with r(t). By inserting u(t) = û(t) + ū(t) into (2.18) we get y(t) = G 0 û + n(t) + G 0 (q)ū (2.19) where the noise sequence n(t) + G 0 (q)ū is uncorrelated with the simulated input signal û(t). Like the direct approach, the two-stage method does not take the complexity of the controller into consideration. The method is valid for any controller, linear or non-linear [Ljung, 1999, p. 440]. A model estimated with the direct approach suffers from a bias if the noise model is poor [Ljung, 1999, p. 433]. For the two-stage method, this is avoided at the cost of increased variance from the extra noise G 0 (q)ū.

25 2.4 Model Validation Model Validation It is important to test the validity of a model. A common validation method is to simulate the output, ŷ(t, Θ), from the measured input and compare it with the measured output, y(t). The model fit, given in %, is calculated by ε(t, ˆΘ) M f = 1 y(t) ȳ 100% (2.20) where ε(t, ˆΘ) is the prediction error from (2.6), the L 2 norm and ȳ is the mean value of y(t). The model fit presents a good and simple measure when comparing different models. The higher model fit, the better model. To be sure that the model captures the dynamics of the system and not disturbances, the validation data needs to be different from the data used when estimating the parameter values. This validation method is called cross-validation [Ljung, 1999, p. 500]. When cross-validating a model, it happens that a more complex model of a higher order has similar or worse model fit than a simpler one. This could be due to a phenomenon called overfit [Ljung, 1999, p. 501]. A model which suffers from overfit has taken the noise of the estimation data into too much account when minimizing the loss function. If the noise is uncorrelated, the noise of a fresh data set is different compared to the noise of the estimation data. Therefore the more complex model will have similar, or potentially worse, model fit than the simple model. If the models were validated using the same data, the model fit would increase with higher order of the model. Another way to compare different models is Akaike s information criterion (aic) which provides a measure of model quality with regard to the order of the model. The model with the smallest aic value is the most accurate according to Akaike s theory [Ljung, 1999, p. 221]. 2.5 Experiment Design When conducting experiments and collecting data to be used for system identification, it is important that the system is persistently excited. A system is said to be persistently excited when the input frequency content is rich enough for the application. Ideally, all frequencies should be equally represented. However this might be unwise due to multiple reasons such as restrictions in maximum output levels. Instead the input should be chosen such that the fastest dynamics of the system to be identified is excited [Ljung, 1999, p ]. One popular way to perform an experiment is to make a step, where the input signal switches instantly from one value to another. The spectrum of an ideal step contains all frequencies and is thus a good start for most data collections. However, for some applications, a step might be difficult or dangerous to use as input. For example, if the input is set manually by a joystick, an ideal step might be too difficult to perform. The change might be step-like enough for some applications while for other it simply is not enough. For these systems another

26 12 2 System Identification approach needs to be taken. A changing signal with as wide spectrum as possible is desired. A popular choice is the chirp signal, also known as a swept sinusoid. A chirp signal is a sine with a continuously increasing frequency over time [Ljung, 1999, p. 423]. This is a commonly suggested input signal when conducting flight experiments to be used for system identification [Mettler, 2003, p. 42].

27 3 The Quadrotor System This chapter presents the quadrotor used in this thesis and a brief explanation to how the maneuverability of a quadrotor is produced. Thereafter, a short description of the hardware and the measured signals is given. The chapter ends with a short presentation of the development of the platform done during this thesis. The quadrotor studied is an X-mounted quadrotor. X-mounted means that the rotors are mounted like an X, with the forward direction in-between two rotors. The other commonly used mount is the -mount, which has one rotor in the forward direction. In other words, the -configuration is rotated 45 compared to the X-mounted quadrotor. To be able to fly, the quadrotor is controlled with a commercial controller for miniature rotor crafts. There is also an avionics solution, developed in the previous master thesis [Blomberg, 2015] with continued development during this thesis. The system is approximately 60 cm across from motor to motor and weighs approximately 1.5 kg. The rotors have two 12 cm long blades each. The avionics solution samples three different sensors, an accelerometer, a gyroscope and a magnetometer, at 50Hz, all which are part of the same inertial measurement unit (imu). All sensors are three dimensional, resulting in nine degrees of freedom for the imu. In addition the avionics solution samples the four pulse width modulated (pwm) signals sent to the motors from the controller as well as the signals sent from the pilot s joystick. In Figure 3.1 a system overview is displayed with the different signals used in this thesis. 3.1 Motion of a Quadrotor To fully describe the movement of the quadrotor, two coordinate systems are convenient to introduce. The first system, denoted E, is an earth-fixed system with its origin in the take-off position of the quadrotor. The axes are denoted 13

28 14 3 The Quadrotor System δ F u Ω T Motors Rotors Rigid body mechanics a, ω, m G G c Avionics solution Figure 3.1: Box diagram of the quadrotor system used in this thesis. The signals measured are the reference signal from the RC-joystick (δ), the four pwm signals sent to the motors (u), the linear accelerations of the quadrotor (a), the angular velocities (ω) and the magnetic field (m). The block labeled F is an unknown flight controller while G is the dynamics of the quadrotor. The signal Ω is the speed of the motors and T is the thrust force from each motor. Ω and T are not measured. (X E, Y E, Z E ) with corresponding unit vectors (e E X, ee Y, ee Z ). The earth-fixed system is a North-East-Down system with X E pointing towards the geographic north, Y E east and Z E towards the center of the earth. The second system is a body-fixed system, denoted B, with its origin fixed on the quadrotor. The body-fixed axes are denoted X B, Y B and Z B respectively with X B pointing forward, Y B right and Z B down. The body-fixed axes are illustrated in Figure 3.2. The unit vectors for the body-fixed system are (e B X, eb Y, eb Z ). These coordinate frames are very common when describing the dynamics of a multirotor system and no further details will be given here, see Lundmark and Lindblom [2015] for more details. The origin of the motions of a multirotor platform is frequently described in literature. Thus, only a short summary is given here, for a more detailed description see Lundmark and Lindblom [2015]. When a rotor is rotating at constant velocity it results in a force, called thrust, and a torque, called reaction torque. The rotors are fixed and therefore the thrust from the rotor is always acting in the negative Z B direction, and the torque is exerted in the opposite direction of the rotation. Figure 3.2 shows a schematic image of the quadrotor with numbered motors and their corresponding rotation directions. The maneuverability of a quadrotor is achieved due to thrust differences between rotors. If the thrust from each rotor is equal, the total reaction torque is zero. This is because two rotors are rotating clock-wise and two counter clockwise. The collective thrust acting on the center of gravity is the sum of all rotors individual thrusts. If the collective thrust equals the force of gravity the quadrotor is in a steady-state called hover. There are three different angular maneuvers in flight mechanics called roll, pitch and yaw with angles denoted φ, θ and ψ and rates p, q and r respectively. The roll angle is the angle around X B while the pitch angle is the angle around

29 3.2 Measured Signals l X B Z B Y B 4 3 Figure 3.2: Rotor numbering convention of the quadrotor and their corresponding rotation directions seen from above. X B indicates the forward direction. the Y B axis. The yaw angle is the angle around the Z B axis. To achieve a right roll, the thrust from motor pair 1 and 4 is increased. To maintain the altitude (height) of the quadrotor the thrust from the opposite pair (2 and 3) needs to be decreased. The net torque is still zero since the speed for each pair are increased/decreased the same amount, motor 1 counters the torque produced by motor 4 and motor 2 counters the torque from motor 3. The pitch is changed accordingly but with the motor pairs 1, 2 and 3, 4. The yaw is a more complex motion to describe and grasp. The yaw rate is produced by letting the resulting reacting torque from the motors be non-zero. A clock-wise yaw rate, seen from above, is achieved by increasing the speed for the counter clock-wise motors (2,4). The reacting torque (that is exerted in the opposite direction of the rotor rotation) produces a net torque that rotates the quadrotor and increases the yaw rate. 3.2 Measured Signals In Figure 3.1, five measured signals are introduced: δ, u, a, ω and m. These signals are sampled by the avionics solution and δ is the joystick position, u is the control signal to a motor and a, ω and m the measurements from the imu. The figure is a simplified illustration of the quadrotor. In reality, u consists of four different signals, one for each motor, while δ consists of four joystick positions, one for collective thrust, roll rate, pitch rate and yaw rate, respectively. The control signals to the motors are produced as pulse width modulated (pwm) signals. The value of a pwm signal is decided by its duty-cycle. A dutycycle is the time for which the signal is high [Barr, 2001]. The sampled pwm signals presented in this thesis are normalized and given as a number from zero to one, where zero means low all the time and one means high all the time. We

30 16 3 The Quadrotor System introduce the notation u i [0, 1], i = {1, 2, 3, 4} which denotes the pwm signal sent to motor i according to the numbering convention in Figure 3.2. From the reasoning in Section 3.1, regarding how the different motions are produced, we introduce u T = u 1 + u 2 + u 3 + u 4 [0, 1] 4 u φ = u 1 u 2 u 3 + u 4 [ 0.5, 0.5] 4 u θ = u 1 + u 2 u 3 u 4 [ 0.5, 0.5] 4 u ψ = u 1 + u 2 u 3 + u 4 [ 0.5, 0.5] 4 (3.1) as the control signal for collective thrust, roll rate, pitch rate and yaw rate, respectively. The expressions for the control signals are divided with four to keep the original magnitude of the control signals. However, the sampled pwm signals suffer from measurement noise. Further, we introduce δ T, δ φ, δ θ and δ ψ for the joystick position controlling collective thrust, roll rate, pitch rate and yaw rate, respectively. These signals are read from a digital serial bus and are thus more precisely measured than the pwm signals. The values for the different joystick positions are given in the range [0, 1]. The trends are removed from the pwm signals and the joystick positions, if not otherwise specified. 3.3 Platform Development During this thesis the platform was developed further. The accelerometer suffered from lots of noise originating from the rotors. Unfortunately, the noise spectrum was evenly distributed, without any indications to the frequencies of the motors. A plot of the accelerometer data from an early flight is shown in Figure 3.3a with the periodogram for the flight in Figure 3.3b. A first improvement was to fixate the imu with a small piece of Blu-Tack. The simple solution significantly reduced the magnitude of the noise. To reduce the noise further, a vibrations damping platform was developed using a 3D-printer and small pieces of a vibration damping material called sylomer SR-11. The static range of use for the sylomer SR-11 is 0.011N/mm 2 and by weighing the avionics the cross sectional area for which the pressure matches the static range of the sylomer can be calculated. The avionics weighs approximately 220g which results in a 0.5cm 2 cross sectional area for each of the four sylomer pieces. The pieces with area 0.5cm 2 were quite flexible and was considered too weak for balancing the platform. Therefore more sturdy pieces were mounted, with approximately 1.5cm 2 cross sectional areas. The noise was significantly reduced for the accelerometer data in the X B and Y B components as well as the yaw rate gyro measurement, but without any significant impact on the Z B component for the accelerometer or the roll and pitch rates. In Figure 3.4 a 10 second yaw rate measurement at hover is shown. See Appendix B for plots comparing the rest

31 3.3 Platform Development 17 Acceleration yacc,z (g) db/hz Time (s) Frequency (Hz) (a) Time plot of the flights. Motors are stared at time t = 190s. (b) Periodogram calculated for time t = [220, 265]s. Figure 3.3: Accelerometer data from an early flight. Notice how the noise of the accelerometer increases when the motors are started. std = 8.5 std = r ( /s) 0 r ( /s) Time (s) Time (s) (a) Yaw rate before the new platform. (b) Yaw rate after the new platform. Figure 3.4: Yaw rate measurements from a ten second long hovering flight before and after the vibration dampening platform was constructed. of the gyroscope and accelerometer measurements before and after the new platform. From those plots it is clear that the horizontal vibrations are reduced. The standard deviation (std) of a x, a y and r is significantly smaller for the data gathered after the new platform was mounted. However the vertical vibrations are not affected very much. The spectrum for the accelerometer data remains evenly distributed as previously with no indication of the rotor frequencies, however the magnitudes are reduced with approximately 10dB for the horizontal accelerations. This shows that the smaller theoretical area was needed to cancel out the vertical noise, while the larger pieces were good enough for the horizontal vibrations. Further the pwm signals were quite noisy to begin with. The pwm signals are sampled by a microprocessor, sampling the signals at a higher rate than 50Hz which is the sampling frequency of the avionics solution. In between the samples,

32 18 3 The Quadrotor System u3 0.6 u Time (s) (a) pwm signal before recursive mean implementation. std = Time (s) (b) pwm signal after recursive mean implementation. std = Figure 3.5: pwm signals from motor three during ten seconds of hover flight. Notice how the std is significantly smaller for the right plot. The trends are not removed. the previously read pwm signals were discarded and only the latest commands to the motors at each sample time were logged. This caused the pwm signals to appear quite noisy. To improve the measurements of the pwm signals a recursive mean algorithm was implemented. The recursive mean algorithm is given by u(k + 1) = k k + 1 u(k) + x k + 1 (3.2) where u is the mean value of the signal, x is the latest read signal and k a counter which is incremented 1 for each value read. When the mean value, u, is read by the avionics solution, the counter, k, is set to zero. A pwm signal from motor three before and after the mean value implementation was made is shown in Figure 3.5. Each plot is is a ten seconds interval for motor three when the quadrotor is in hover. As can be seen in the plots, the std of the pwm signals were reduced.

33 4 Physical Model This chapter contains the derivation of the physical model that is used as structure for the grey-box system identification. The chapter begins with a rigid body approach using Newton and Euler laws of motion. Then the electrical motors and the relation between input voltage and thrust force from the rotors are studied. The non-linear model is linearized and the linear model that will be used as model structure is presented. Finally, the model is split into four sub-models describing the dynamics of the vertical velocity, the roll rate, the pitch rate and the yaw rate. 4.1 States The quadrotor s absolute position can be defined in the earth-fixed coordinate system, E, since it acts as an inertial frame. The position of the origin of the body-fixed system, B, in E is denoted as ξ E = (x, y, z) T. (4.1) The orientation of B in relation to E is described by the vector η E = (φ, θ, ψ) T (4.2) where φ, θ and ψ are the roll, pitch and yaw angle, respectively. The linear velocities of the quadrotor in B are denoted as and the angular rates of the quadrotor are denoted as ν B = (u, v, w) T (4.3) ω B = (p, q, r) T, (4.4) 19

34 20 4 Physical Model respectively. For readability, the superscript E and B will be dropped further on, unless there is an explicit need to use them. Using these vectors we introduce the state vector X = (ξ, ν, η, ω) T = (x, y, z, u, v, w, φ, θ, ψ, p, q, r) T (4.5) that describes the motion of the quadrotor. 4.2 Kinetics To transform a vector expressed in the earth-fixed frame to a vector expressed in the body-fixed frame, the vector is transformed using the transformation matrix c θ c ψ c θ s ψ s θ C B E = s φ s θ c ψ c φ s ψ s φ s θ s ψ + c φ c ψ s φ c θ (4.6) s φ s θ s ψ + c φ c ψ c φ s θ s ψ s φ c ψ c φ c θ where c x and s x is short notation for cos(x) and sin(x), respectively. The transformation from the body-fixed frame to the earth-fixed frame, C E B, is given by the inverse of C B E which, since it is orthogonal, equals the transpose C E B = ( ) C B 1 ( ) c E = C B T θ c ψ s φ s θ c ψ c φ s ψ c φ s θ c ψ + s φ s ψ E = c θ s ψ s φ s θ s ψ + c φ c ψ c φ s θ s ψ s φ c ψ. (4.7) s θ s φ c θ c φ c θ The transformation matrix for the angular rates is given by ω E = C E B,rot ωb (4.8) 1 s φ t θ c φ t θ C E B,rot = 0 c φ s φ 0 s φ c θ c φ c θ (4.9) where t x is the short notation for tan(x). For full derivation of the matrices, see Lundmark and Lindblom [2015]. Using the linear rotational matrix from (4.7) the time derivative of the earthfixed position of the quadrotor from (4.1) is given as ξ E = ν E = C E B ν B (4.10) and using the angular rotational matrix defined in (4.9) the time derivative of η E from (4.2) is η E = ω E = C E B,rot ωb. (4.11) Using the Newton-Euler formalism as in Bouabdallah et al. [2004] the force, F, and torque, M, equations for the quadrotor can be expressed as F =m ν + ω mν (4.12) M =I ω + ω I ω (4.13) where I is the inertia matrix and m is the mass of the quadrotor.

35 4.2 Kinetics Force Equations The forces acting on the quadrotor are the gravity, and the thrust from the rotors, F g = mge E z = mgc B E eb z, (4.14) F T = T e B z. (4.15) Hoffmann et al. [2007] suggest that the quadrotor is affected by a drag force along the velocity vector F D = Dν. (4.16) In [Mettler, 2003, p. 91], the same factor is suggested for a miniature helicopter. The total force equation is thus F = F g + F T + F D. (4.17) The collective thrust is the sum of the thrust forces 4 T = T i (4.18) from the four motors. Using (4.14)-(4.18) in (4.12) we get i=1 ν = ω ν D m ν + gcb E eb z 1 m T eb z. (4.19) Inserting (4.3), (4.4) and (4.6) in (4.19) and assuming that the drag coefficient, D, is diagonal, ν becomes u rv qw D x u s θ g v = pw ru D y v + s φ c θ g ẇ qu pv D z w + c φ c θ g m 1 T. (4.20) Torque Equations The torque equation (4.13) can be rewritten as an expression for the time derivative of ω as ω = I 1 M I 1 (ω I ω) (4.21) where the inertia matrix, I, is defined as I xx I xy I xz I = I yx I yy I yz I zx I zy I zz (4.22) where I xy = I yx, I xz = I zx and I yz = I zy. The symmetry of the quadrotor suggests that I xy = I xz = I yz = 0 hence I xx 0 0 I = 0 I yy 0. (4.23) 0 0 I zz

36 22 4 Physical Model Calculating (4.21) using (4.4) and (4.23) it becomes 1 ( ) 1 I 0 0 xx I Iyy I xx zz qr 1 ω = 0 Iyy M + 1 I (I yy zz I xx pr ( ) 1 Ixx I Ixx I zz yy pq (4.24) In order to fully describe ω, the total torque exerted on the quadrotor, M, must be derived. There are several torques acting on the quadrotor such as induced torque from the thrust force and gravity. In addition there is a reaction torque produced by the rotation of the rotors, an inertial counter torque produced by a change in the rotors angular velocity as well as a gyroscopic torque which is exerted on the body by simultaneous spin and precession motion [Lundmark and Lindblom, 2015]. In this thesis the torque from the gyroscopic rotor effects and the inertial counter torque are neglected. Using the torque expression from Lundmark and Lindblom [2015] assuming that the center of gravity coincide with the origin of the body-fixed frame and neglecting the gyroscopic rotor effects, the total torque becomes l T 2 φ M = T 2 θ k Q T ψ l (4.25) where k Q is a reaction torque coefficient and l the length of the quadrotor arms. The torque is manipulated by different combination of thrust for the different motors according to Section 3.1 and the different thrust combinations can be expressed by T φ = T 1 T 2 T 3 + T 4 (4.26a) T θ = T 1 + T 2 T 3 T 4 (4.26b) T ψ = T 1 + T 2 T 3 + T 4 (4.26c) where T i is the thrust force from rotor i. In [Mettler, 2003, p. 91], a similar drag coefficient for the angular velocities as for the linear velocities (see Section 4.2.1) is suggested. Therefore, an unknown drag coefficient for the rates is added. (4.27) together with (4.25) in (4.24), yield ṗ D φ 0 0 ω D = 0 D θ D ω (4.27) ψ ( (Iyy ) ) 1 I I xx zz qr + l T 2 φ D φ p ) ω = q = 1 (I zz I xx ) pr + l T 2 θ D θ q. (4.28) ṙ (( ) ) 1 I Ixx I zz yy pq + kq T ψ Dψ r I yy (

37 4.3 Motor and Rotor Dynamics 23 The geometry of the quadrotor platform suggests that the inertia around the X B and the Y B axis is similar. We therefore assume I xx = I yy and assign I = I yy I zz I xx = I xx I zz I yy. With this approximation (4.28) becomes Iqr + l T 2I φ D φ p 4.3 Motor and Rotor Dynamics l ω = Ipr + T 2I θ D θ q. (4.29) 1 I k zz Q T ψ D ψ r So far the physical model only contains information for the rigid body mechanics of the quadrotor, from thrust to motion. For the dynamics from motor voltage to thrust, a model for the electrical motors and the thrust produced by a rotor is needed. An electrical motor often is modelled as a first order system Ω i = 1 τ Ω i + K τ i u i (4.30) where Ω i is the angular speed of the motor and u i is the input voltage [Fitzgerald et al., 2014]. The constants K and τ is the gain and time constant of the system respectively. The motor constants, K and τ, is assumed to be equal for the four motors. In some studies, such as Bergman and Ekström [2014], the time constant, τ, for the rotor dynamics is shown different for an accelerating and a decelerating rotor due to various aerodynamic effects and because there are no active brakes for the motors. This behaviour will be neglected in this thesis mainly for the simplicity of the model but also since the difference is more prominent for large aggressive differences in angular speed for the motors compared to smaller changes [Lundmark and Lindblom, 2015]. Since the model is developed with regard to future control design, the simplicity of the model is more important than to accurately describe minor non-linearities such as the different motor constants for small speed changes. According to Pounds et al. [2004] the thrust force, T i, from rotor i can be modelled as T i = C T ρa(ω i R) 2 (4.31) where C T is a thrust coefficient, ρ the air density, A the area of the rotor, Ω i the angular speed of the rotor and R the radius of the rotor. We rewrite the expression T i = k r Ω 2 i (4.32) where k r = C T ρar 2 is a rotor specific constant which is assumed to be equal for all four rotors.

Linköping University Electronic Press

Linköping University Electronic Press Linköping University Electronic Press Report Simulation Model of a 2 Degrees of Freedom Industrial Manipulator Patrik Axelsson Series: LiTH-ISY-R, ISSN 400-3902, No. 3020 ISRN: LiTH-ISY-R-3020 Available

More information

An efficient implementation of gradient and Hessian calculations of the coefficients of the characteristic polynomial of I XY

An efficient implementation of gradient and Hessian calculations of the coefficients of the characteristic polynomial of I XY Technical report from Automatic Control at Linköpings universitet An efficient implementation of gradient and Hessian calculations of the coefficients of the characteristic polynomial of I XY Daniel Ankelhed

More information

Robot Dynamics - Rotary Wing UAS: Control of a Quadrotor

Robot Dynamics - Rotary Wing UAS: Control of a Quadrotor Robot Dynamics Rotary Wing AS: Control of a Quadrotor 5-85- V Marco Hutter, Roland Siegwart and Thomas Stastny Robot Dynamics - Rotary Wing AS: Control of a Quadrotor 7..6 Contents Rotary Wing AS. Introduction

More information

Institutionen för systemteknik

Institutionen för systemteknik Institutionen för systemteknik Department of Electrical Engineering Examensarbete Modelling and control of a hexarotor UAV Examensarbete utfört i Reglerteknik vid Tekniska högskolan vid Linköpings universitet

More information

LTI Approximations of Slightly Nonlinear Systems: Some Intriguing Examples

LTI Approximations of Slightly Nonlinear Systems: Some Intriguing Examples LTI Approximations of Slightly Nonlinear Systems: Some Intriguing Examples Martin Enqvist, Lennart Ljung Division of Automatic Control Department of Electrical Engineering Linköpings universitet, SE-581

More information

Quadcopter Dynamics 1

Quadcopter Dynamics 1 Quadcopter Dynamics 1 Bréguet Richet Gyroplane No. 1 1907 Brothers Louis Bréguet and Jacques Bréguet Guidance of Professor Charles Richet The first flight demonstration of Gyroplane No. 1 with no control

More information

System analysis of a diesel engine with VGT and EGR

System analysis of a diesel engine with VGT and EGR System analysis of a diesel engine with VGT and EGR Master s thesis performed in Vehicular Systems by Thomas Johansson Reg nr: LiTH-ISY-EX- -5/3714- -SE 9th October 25 System analysis of a diesel engine

More information

Chapter 4 The Equations of Motion

Chapter 4 The Equations of Motion Chapter 4 The Equations of Motion Flight Mechanics and Control AEM 4303 Bérénice Mettler University of Minnesota Feb. 20-27, 2013 (v. 2/26/13) Bérénice Mettler (University of Minnesota) Chapter 4 The Equations

More information

Lecture AC-1. Aircraft Dynamics. Copy right 2003 by Jon at h an H ow

Lecture AC-1. Aircraft Dynamics. Copy right 2003 by Jon at h an H ow Lecture AC-1 Aircraft Dynamics Copy right 23 by Jon at h an H ow 1 Spring 23 16.61 AC 1 2 Aircraft Dynamics First note that it is possible to develop a very good approximation of a key motion of an aircraft

More information

Block diagonalization of matrix-valued sum-of-squares programs

Block diagonalization of matrix-valued sum-of-squares programs Block diagonalization of matrix-valued sum-of-squares programs Johan Löfberg Division of Automatic Control Department of Electrical Engineering Linköpings universitet, SE-581 83 Linköping, Sweden WWW:

More information

Non-linear Dynamic Inversion

Non-linear Dynamic Inversion Control of Unmanned Aerial Vehicles using Non-linear Dynamic Inversion Master s Thesis Division of Automatic Control Department of Electrical Engineering Linköping University Mia Karlsson LiTH-ISY-EX-3-

More information

CS491/691: Introduction to Aerial Robotics

CS491/691: Introduction to Aerial Robotics CS491/691: Introduction to Aerial Robotics Topic: Midterm Preparation Dr. Kostas Alexis (CSE) Areas of Focus Coordinate system transformations (CST) MAV Dynamics (MAVD) Navigation Sensors (NS) State Estimation

More information

Institutionen för systemteknik

Institutionen för systemteknik Institutionen för systemteknik Department of Electrical Engineering Examensarbete Multiple Platform Bias Error Estimation Examensarbete utfört i Reglerteknik vid Tekniska högskolan i Linköping av Åsa Wiklund

More information

Dynamics and Control of Rotorcraft

Dynamics and Control of Rotorcraft Dynamics and Control of Rotorcraft Helicopter Aerodynamics and Dynamics Abhishek Department of Aerospace Engineering Indian Institute of Technology, Kanpur February 3, 2018 Overview Flight Dynamics Model

More information

Mathematical Modelling and Dynamics Analysis of Flat Multirotor Configurations

Mathematical Modelling and Dynamics Analysis of Flat Multirotor Configurations Mathematical Modelling and Dynamics Analysis of Flat Multirotor Configurations DENIS KOTARSKI, Department of Mechanical Engineering, Karlovac University of Applied Sciences, J.J. Strossmayera 9, Karlovac,

More information

Modelling of Opposed Lateral and Longitudinal Tilting Dual-Fan Unmanned Aerial Vehicle

Modelling of Opposed Lateral and Longitudinal Tilting Dual-Fan Unmanned Aerial Vehicle Modelling of Opposed Lateral and Longitudinal Tilting Dual-Fan Unmanned Aerial Vehicle N. Amiri A. Ramirez-Serrano R. Davies Electrical Engineering Department, University of Calgary, Canada (e-mail: namiri@ucalgary.ca).

More information

Quadrotor Modeling and Control

Quadrotor Modeling and Control 16-311 Introduction to Robotics Guest Lecture on Aerial Robotics Quadrotor Modeling and Control Nathan Michael February 05, 2014 Lecture Outline Modeling: Dynamic model from first principles Propeller

More information

Investigation of the Dynamics and Modeling of a Triangular Quadrotor Configuration

Investigation of the Dynamics and Modeling of a Triangular Quadrotor Configuration Investigation of the Dynamics and Modeling of a Triangular Quadrotor Configuration TONI AXELSSON Master s Thesis at Aerospace Engineering Supervisor: Arne Karlsson Examiner: Arne Karlsson ISSN 1651-7660

More information

Institutionen för systemteknik

Institutionen för systemteknik Institutionen för systemteknik Department of Electrical Engineering Examensarbete Modeling and Identification of the Gantry-Tau Parallel Kinematic Machine Examensarbete utfört i Reglerteknik vid Tekniska

More information

Rotational Motion. Chapter 4. P. J. Grandinetti. Sep. 1, Chem P. J. Grandinetti (Chem. 4300) Rotational Motion Sep.

Rotational Motion. Chapter 4. P. J. Grandinetti. Sep. 1, Chem P. J. Grandinetti (Chem. 4300) Rotational Motion Sep. Rotational Motion Chapter 4 P. J. Grandinetti Chem. 4300 Sep. 1, 2017 P. J. Grandinetti (Chem. 4300) Rotational Motion Sep. 1, 2017 1 / 76 Angular Momentum The angular momentum of a particle with respect

More information

Robot Control Basics CS 685

Robot Control Basics CS 685 Robot Control Basics CS 685 Control basics Use some concepts from control theory to understand and learn how to control robots Control Theory general field studies control and understanding of behavior

More information

Position Estimation and Modeling of a Flexible Industrial Robot

Position Estimation and Modeling of a Flexible Industrial Robot Position Estimation and Modeling of a Flexible Industrial Robot Rickard Karlsson, Mikael Norrlöf, Division of Automatic Control Department of Electrical Engineering Linköpings universitet, SE-581 83 Linköping,

More information

A Comparison of Closed-Loop Performance of Multirotor Configurations Using Non-Linear Dynamic Inversion Control

A Comparison of Closed-Loop Performance of Multirotor Configurations Using Non-Linear Dynamic Inversion Control Aerospace 2015, 2, 325-352; doi:10.3390/aerospace2020325 OPEN ACCESS aerospace ISSN 2226-4310 www.mdpi.com/journal/aerospace Article A Comparison of Closed-Loop Performance of Multirotor Configurations

More information

The PVTOL Aircraft. 2.1 Introduction

The PVTOL Aircraft. 2.1 Introduction 2 The PVTOL Aircraft 2.1 Introduction We introduce in this chapter the well-known Planar Vertical Take-Off and Landing (PVTOL) aircraft problem. The PVTOL represents a challenging nonlinear systems control

More information

Properties and approximations of some matrix variate probability density functions

Properties and approximations of some matrix variate probability density functions Technical report from Automatic Control at Linköpings universitet Properties and approximations of some matrix variate probability density functions Karl Granström, Umut Orguner Division of Automatic Control

More information

CHAPTER 1. Introduction

CHAPTER 1. Introduction CHAPTER 1 Introduction Linear geometric control theory was initiated in the beginning of the 1970 s, see for example, [1, 7]. A good summary of the subject is the book by Wonham [17]. The term geometric

More information

In this section of notes, we look at the calculation of forces and torques for a manipulator in two settings:

In this section of notes, we look at the calculation of forces and torques for a manipulator in two settings: Introduction Up to this point we have considered only the kinematics of a manipulator. That is, only the specification of motion without regard to the forces and torques required to cause motion In this

More information

Artificial Intelligence & Neuro Cognitive Systems Fakultät für Informatik. Robot Dynamics. Dr.-Ing. John Nassour J.

Artificial Intelligence & Neuro Cognitive Systems Fakultät für Informatik. Robot Dynamics. Dr.-Ing. John Nassour J. Artificial Intelligence & Neuro Cognitive Systems Fakultät für Informatik Robot Dynamics Dr.-Ing. John Nassour 25.1.218 J.Nassour 1 Introduction Dynamics concerns the motion of bodies Includes Kinematics

More information

EL1820 Modeling of Dynamical Systems

EL1820 Modeling of Dynamical Systems EL1820 Modeling of Dynamical Systems Lecture 9 - Parameter estimation in linear models Model structures Parameter estimation via prediction error minimization Properties of the estimate: bias and variance

More information

Manipulator Dynamics 2. Instructor: Jacob Rosen Advanced Robotic - MAE 263D - Department of Mechanical & Aerospace Engineering - UCLA

Manipulator Dynamics 2. Instructor: Jacob Rosen Advanced Robotic - MAE 263D - Department of Mechanical & Aerospace Engineering - UCLA Manipulator Dynamics 2 Forward Dynamics Problem Given: Joint torques and links geometry, mass, inertia, friction Compute: Angular acceleration of the links (solve differential equations) Solution Dynamic

More information

Institutionen för systemteknik

Institutionen för systemteknik Institutionen för systemteknik Department of Electrical Engineering Examensarbete Dynamical Analysis and System Identification of the Gantry-Tau Parallel Manipulator Examensarbete utfört i Reglerteknik

More information

ENHANCED PROPORTIONAL-DERIVATIVE CONTROL OF A MICRO QUADCOPTER

ENHANCED PROPORTIONAL-DERIVATIVE CONTROL OF A MICRO QUADCOPTER ENHANCED PROPORTIONAL-DERIVATIVE CONTROL OF A MICRO QUADCOPTER Norman L. Johnson and Kam K. Leang Department of Mechanical Engineering University of Nevada, Reno Reno, Nevada 897-312, USA ABSTRACT This

More information

Gain Scheduled Missile Control Using Robust Loop Shaping Examensarbete utfört i Reglerteknik vid Tekniska Högskolan i Linköping av Henrik Johansson Reg nr: LiTH-ISY-EX-3291-22 Gain Scheduled Missile Control

More information

Nonlinear and Neural Network-based Control of a Small Four-Rotor Aerial Robot

Nonlinear and Neural Network-based Control of a Small Four-Rotor Aerial Robot Nonlinear and Neural Network-based Control of a Small Four-Rotor Aerial Robot Holger Voos Abstract Small four-rotor aerial robots, so called quadrotor UAVs, have an enormous potential for all kind of neararea

More information

Institutionen för systemteknik

Institutionen för systemteknik Institutionen för systemteknik Department of Electrical Engineering Examensarbete Load flow control and optimization of Banverket s 132 kv 16 2/3 Hz high voltage grid Examensarbete utfört i Reglerteknik

More information

Institutionen för systemteknik

Institutionen för systemteknik main: 2005-9-5 9:41 1(1) Institutionen för systemteknik Department of Electrical Engineering Examensarbete A Tracking and Collision Warning System for Maritime Applications Examensarbete utfört i Reglerteknik

More information

Multi Rotor Scalability

Multi Rotor Scalability Multi Rotor Scalability With the rapid growth in popularity of quad copters and drones in general, there has been a small group of enthusiasts who propose full scale quad copter designs (usable payload

More information

On Consistency of Closed-loop Subspace Identifictaion with Innovation Estimation

On Consistency of Closed-loop Subspace Identifictaion with Innovation Estimation Technical report from Automatic Control at Linköpings universitet On Consistency of Closed-loop Subspace Identictaion with Innovation Estimation Weilu Lin, S Joe Qin, Lennart Ljung Division of Automatic

More information

Institutionen för systemteknik

Institutionen för systemteknik Institutionen för systemteknik Department of Electrical Engineering Examensarbete Estimation of Inter-cell Interference in 3G Communication Systems Examensarbete utfört i Reglerteknik vid Tekniska högskolan

More information

On Indirect Input Measurements

On Indirect Input Measurements Technical report from Automatic Control at Linköpings universitet On Indirect Input Measurements Jonas Linder, Martin Enqvist Division of Automatic Control E-mail: jonas.linder@liu.se, maren@isy.liu.se

More information

Robust Control Of A Flexible Manipulator Arm: A Benchmark Problem

Robust Control Of A Flexible Manipulator Arm: A Benchmark Problem Robust Control Of A Flexible Manipulator Arm: A Benchmark Problem Stig Moberg, Jonas Öhr Division of Automatic Control Department of Electrical Engineering Linköpings universitet, SE-581 83 Linköping,

More information

Adaptive Trim and Trajectory Following for a Tilt-Rotor Tricopter Ahmad Ansari, Anna Prach, and Dennis S. Bernstein

Adaptive Trim and Trajectory Following for a Tilt-Rotor Tricopter Ahmad Ansari, Anna Prach, and Dennis S. Bernstein 7 American Control Conference Sheraton Seattle Hotel May 4 6, 7, Seattle, USA Adaptive Trim and Trajectory Following for a Tilt-Rotor Tricopter Ahmad Ansari, Anna Prach, and Dennis S. Bernstein Abstract

More information

Lane departure detection for improved road geometry estimation

Lane departure detection for improved road geometry estimation Lane departure detection for improved road geometry estimation Thomas B. Schön, Andreas Eidehall, Fredrik Gustafsson Division of Automatic Control Department of Electrical Engineering Linköpings universitet,

More information

6. 3D Kinematics DE2-EA 2.1: M4DE. Dr Connor Myant

6. 3D Kinematics DE2-EA 2.1: M4DE. Dr Connor Myant DE2-EA 2.1: M4DE Dr Connor Myant 6. 3D Kinematics Comments and corrections to connor.myant@imperial.ac.uk Lecture resources may be found on Blackboard and at http://connormyant.com Contents Three-Dimensional

More information

Flight Dynamics and Control

Flight Dynamics and Control Flight Dynamics and Control Lecture 1: Introduction G. Dimitriadis University of Liege Reference material Lecture Notes Flight Dynamics Principles, M.V. Cook, Arnold, 1997 Fundamentals of Airplane Flight

More information

AB-267 DYNAMICS & CONTROL OF FLEXIBLE AIRCRAFT

AB-267 DYNAMICS & CONTROL OF FLEXIBLE AIRCRAFT FLÁIO SILESTRE DYNAMICS & CONTROL OF FLEXIBLE AIRCRAFT LECTURE NOTES LAGRANGIAN MECHANICS APPLIED TO RIGID-BODY DYNAMICS IMAGE CREDITS: BOEING FLÁIO SILESTRE Introduction Lagrangian Mechanics shall be

More information

Application of state observers in attitude estimation using low-cost sensors

Application of state observers in attitude estimation using low-cost sensors Application of state observers in attitude estimation using low-cost sensors Martin Řezáč Czech Technical University in Prague, Czech Republic March 26, 212 Introduction motivation for inertial estimation

More information

Chapter 2 Review of Linear and Nonlinear Controller Designs

Chapter 2 Review of Linear and Nonlinear Controller Designs Chapter 2 Review of Linear and Nonlinear Controller Designs This Chapter reviews several flight controller designs for unmanned rotorcraft. 1 Flight control systems have been proposed and tested on a wide

More information

MODELING AND IDENTIFICATION OF A MECHANICAL INDUSTRIAL MANIPULATOR 1

MODELING AND IDENTIFICATION OF A MECHANICAL INDUSTRIAL MANIPULATOR 1 Copyright 22 IFAC 15th Triennial World Congress, Barcelona, Spain MODELING AND IDENTIFICATION OF A MECHANICAL INDUSTRIAL MANIPULATOR 1 M. Norrlöf F. Tjärnström M. Östring M. Aberger Department of Electrical

More information

Theory and Practice of Rotor Dynamics Prof. Dr. Rajiv Tiwari Department of Mechanical Engineering Indian Institute of Technology Guwahati

Theory and Practice of Rotor Dynamics Prof. Dr. Rajiv Tiwari Department of Mechanical Engineering Indian Institute of Technology Guwahati Theory and Practice of Rotor Dynamics Prof. Dr. Rajiv Tiwari Department of Mechanical Engineering Indian Institute of Technology Guwahati Module - 2 Simpul Rotors Lecture - 2 Jeffcott Rotor Model In the

More information

Introduction to Haptic Systems

Introduction to Haptic Systems Introduction to Haptic Systems Félix Monasterio-Huelin & Álvaro Gutiérrez & Blanca Larraga October 8, 2018 Contents Contents 1 List of Figures 1 1 Introduction 2 2 DC Motor 3 3 1 DOF DC motor model with

More information

Institutionen för systemteknik

Institutionen för systemteknik Institutionen för systemteknik Department of Electrical Engineering Examensarbete Sensor-less Control of a Permanent Magnet Synchronous Motor Examensarbete utfört i Reglerteknik vid Tekniska högskolan

More information

Robot Dynamics Rotorcrafts: Dynamic Modeling of Rotorcraft & Control

Robot Dynamics Rotorcrafts: Dynamic Modeling of Rotorcraft & Control Robot Dynamics Rotorcrafts: Dynamic Modeling of Rotorcraft & Control 5-85- V Marco Hutter, Roland Siegwart and Thomas Stastny Robot Dynamics - Rotary Wing UAS: Propeller Analysis and Dynamic Modeling 7..5

More information

Institutionen för systemteknik

Institutionen för systemteknik main: 2006-9-19 17:45 1(1) Institutionen för systemteknik Department of Electrical Engineering Examensarbete Design and Implementation of a Test Rig for a Gyro Stabilized Camera System Examensarbete utfört

More information

Nonlinear Landing Control for Quadrotor UAVs

Nonlinear Landing Control for Quadrotor UAVs Nonlinear Landing Control for Quadrotor UAVs Holger Voos University of Applied Sciences Ravensburg-Weingarten, Mobile Robotics Lab, D-88241 Weingarten Abstract. Quadrotor UAVs are one of the most preferred

More information

General Physics I. Lecture 10: Rolling Motion and Angular Momentum.

General Physics I. Lecture 10: Rolling Motion and Angular Momentum. General Physics I Lecture 10: Rolling Motion and Angular Momentum Prof. WAN, Xin (万歆) 万歆 ) xinwan@zju.edu.cn http://zimp.zju.edu.cn/~xinwan/ Outline Rolling motion of a rigid object: center-of-mass motion

More information

Lab #4 - Gyroscopic Motion of a Rigid Body

Lab #4 - Gyroscopic Motion of a Rigid Body Lab #4 - Gyroscopic Motion of a Rigid Body Last Updated: April 6, 2007 INTRODUCTION Gyroscope is a word used to describe a rigid body, usually with symmetry about an axis, that has a comparatively large

More information

Visual Servoing for a Quadrotor UAV in Target Tracking Applications. Marinela Georgieva Popova

Visual Servoing for a Quadrotor UAV in Target Tracking Applications. Marinela Georgieva Popova Visual Servoing for a Quadrotor UAV in Target Tracking Applications by Marinela Georgieva Popova A thesis submitted in conformity with the requirements for the degree of Master of Applied Science Graduate

More information

Detecting external forces on an autonomous lawn mowing robot with inertial, wheel speed and wheel motor current measurements

Detecting external forces on an autonomous lawn mowing robot with inertial, wheel speed and wheel motor current measurements Master of Science Thesis in reglerteknik Department of Electrical Engineering, Linköping University, 2017 Detecting external forces on an autonomous lawn mowing robot with inertial, wheel speed and wheel

More information

Simplified Analytical Model of a Six-Degree-of-Freedom Large-Gap Magnetic Suspension System

Simplified Analytical Model of a Six-Degree-of-Freedom Large-Gap Magnetic Suspension System NASA Technical Memorandum 112868 Simplified Analytical Model of a Six-Degree-of-Freedom Large-Gap Magnetic Suspension System Nelson J. Groom Langley Research Center, Hampton, Virginia June 1997 National

More information

Model Reduction using a Frequency-Limited H 2 -Cost

Model Reduction using a Frequency-Limited H 2 -Cost Technical report from Automatic Control at Linköpings universitet Model Reduction using a Frequency-Limited H 2 -Cost Daniel Petersson, Johan Löfberg Division of Automatic Control E-mail: petersson@isy.liu.se,

More information

c 2016 Jan Willem Vervoorst

c 2016 Jan Willem Vervoorst c 216 Jan Willem Vervoorst A MODULAR SIMULATION ENVIRONMENT FOR THE IMPROVED DYNAMIC SIMULATION OF MULTIROTOR UNMANNED AERIAL VEHICLES BY JAN WILLEM VERVOORST THESIS Submitted in partial fulfillment of

More information

A relaxation of the strangeness index

A relaxation of the strangeness index echnical report from Automatic Control at Linköpings universitet A relaxation of the strangeness index Henrik idefelt, orkel Glad Division of Automatic Control E-mail: tidefelt@isy.liu.se, torkel@isy.liu.se

More information

Department of Physics, Chemistry and Biology

Department of Physics, Chemistry and Biology Department of Physics, Chemistry and Biology Master s Thesis Quantum Chaos On A Curved Surface John Wärnå LiTH-IFM-A-EX-8/7-SE Department of Physics, Chemistry and Biology Linköpings universitet, SE-58

More information

System Identification and H Observer Design for TRMS

System Identification and H Observer Design for TRMS International Journal of Computer and Electrical Engineering, Vol. 5, No. 6, December 3 System Identification and H Observer Design for TRMS Vidya S. Rao, Milind Mukerji, V. I. George, Surekha Kamath,

More information

with Application to Autonomous Vehicles

with Application to Autonomous Vehicles Nonlinear with Application to Autonomous Vehicles (Ph.D. Candidate) C. Silvestre (Supervisor) P. Oliveira (Co-supervisor) Institute for s and Robotics Instituto Superior Técnico Portugal January 2010 Presentation

More information

Circumnavigation with a group of quadrotor helicopters

Circumnavigation with a group of quadrotor helicopters Circumnavigation with a group of quadrotor helicopters JOHANNA ORIHUELA SWARTLING Master s Degree Project Stockholm, Sweden March 2014 XR-EE-RT 2014:007 Abstract The primary goal of this thesis has been

More information

Autonomous Helicopter Flight via Reinforcement Learning

Autonomous Helicopter Flight via Reinforcement Learning Autonomous Helicopter Flight via Reinforcement Learning Authors: Andrew Y. Ng, H. Jin Kim, Michael I. Jordan, Shankar Sastry Presenters: Shiv Ballianda, Jerrolyn Hebert, Shuiwang Ji, Kenley Malveaux, Huy

More information

Quadrotor Modeling and Control for DLO Transportation

Quadrotor Modeling and Control for DLO Transportation Quadrotor Modeling and Control for DLO Transportation Thesis dissertation Advisor: Prof. Manuel Graña Computational Intelligence Group University of the Basque Country (UPV/EHU) Donostia Jun 24, 2016 Abstract

More information

Institutionen för systemteknik

Institutionen för systemteknik Institutionen för systemteknik Department of Electrical Engineering Examensarbete Vehicle Mass and Road Grade Estimation Using Kalman Filter Examensarbete utfört i Fordonssystem vid Tekniska högskolan

More information

Control Systems Lab - SC4070 System Identification and Linearization

Control Systems Lab - SC4070 System Identification and Linearization Control Systems Lab - SC4070 System Identification and Linearization Dr. Manuel Mazo Jr. Delft Center for Systems and Control (TU Delft) m.mazo@tudelft.nl Tel.:015-2788131 TU Delft, February 13, 2015 (slides

More information

MAE 142 Homework #5 Due Friday, March 13, 2009

MAE 142 Homework #5 Due Friday, March 13, 2009 MAE 142 Homework #5 Due Friday, March 13, 2009 Please read through the entire homework set before beginning. Also, please label clearly your answers and summarize your findings as concisely as possible.

More information

1 Kalman Filter Introduction

1 Kalman Filter Introduction 1 Kalman Filter Introduction You should first read Chapter 1 of Stochastic models, estimation, and control: Volume 1 by Peter S. Maybec (available here). 1.1 Explanation of Equations (1-3) and (1-4) Equation

More information

LQ Control of a Two Wheeled Inverted Pendulum Process

LQ Control of a Two Wheeled Inverted Pendulum Process Uppsala University Information Technology Dept. of Systems and Control KN,HN,FS 2000-10 Last rev. September 12, 2017 by HR Reglerteknik II Instruction to the laboratory work LQ Control of a Two Wheeled

More information

LAWS OF GYROSCOPES / CARDANIC GYROSCOPE

LAWS OF GYROSCOPES / CARDANIC GYROSCOPE LAWS OF GYROSCOPES / CARDANC GYROSCOPE PRNCPLE f the axis of rotation of the force-free gyroscope is displaced slightly, a nutation is produced. The relationship between precession frequency or nutation

More information

Institutionen för systemteknik

Institutionen för systemteknik Institutionen för systemteknik Department of Electrical Engineering Examensarbete Intelligent Body Monitoring Examensarbete utfört i Reglerteknik vid Tekniska högskolan vid Linköpings universitet av Rikard

More information

Nonlinear Control of a Quadrotor Micro-UAV using Feedback-Linearization

Nonlinear Control of a Quadrotor Micro-UAV using Feedback-Linearization Proceedings of the 2009 IEEE International Conference on Mechatronics. Malaga, Spain, April 2009. Nonlinear Control of a Quadrotor Micro-UAV using Feedback-Linearization Holger Voos University of Applied

More information

12. Prediction Error Methods (PEM)

12. Prediction Error Methods (PEM) 12. Prediction Error Methods (PEM) EE531 (Semester II, 2010) description optimal prediction Kalman filter statistical results computational aspects 12-1 Description idea: determine the model parameter

More information

Automatic Control II Computer exercise 3. LQG Design

Automatic Control II Computer exercise 3. LQG Design Uppsala University Information Technology Systems and Control HN,FS,KN 2000-10 Last revised by HR August 16, 2017 Automatic Control II Computer exercise 3 LQG Design Preparations: Read Chapters 5 and 9

More information

EECE Adaptive Control

EECE Adaptive Control EECE 574 - Adaptive Control Recursive Identification in Closed-Loop and Adaptive Control Guy Dumont Department of Electrical and Computer Engineering University of British Columbia January 2010 Guy Dumont

More information

Lecture 11 Overview of Flight Dynamics I. Dr. Radhakant Padhi Asst. Professor Dept. of Aerospace Engineering Indian Institute of Science - Bangalore

Lecture 11 Overview of Flight Dynamics I. Dr. Radhakant Padhi Asst. Professor Dept. of Aerospace Engineering Indian Institute of Science - Bangalore Lecture 11 Overview of Flight Dynamics I Dr. Radhakant Padhi Asst. Professor Dept. of Aerospace Engineering Indian Institute of Science - Bangalore Point Mass Dynamics Dr. Radhakant Padhi Asst. Professor

More information

Kinetic Energy of Rolling

Kinetic Energy of Rolling Kinetic Energy of Rolling A solid disk and a hoop (with the same mass and radius) are released from rest and roll down a ramp from a height h. Which one is moving faster at the bottom of the ramp? A. they

More information

Modal Analysis: What it is and is not Gerrit Visser

Modal Analysis: What it is and is not Gerrit Visser Modal Analysis: What it is and is not Gerrit Visser What is a Modal Analysis? What answers do we get out of it? How is it useful? What does it not tell us? In this article, we ll discuss where a modal

More information

MEM04: Rotary Inverted Pendulum

MEM04: Rotary Inverted Pendulum MEM4: Rotary Inverted Pendulum Interdisciplinary Automatic Controls Laboratory - ME/ECE/CHE 389 April 8, 7 Contents Overview. Configure ELVIS and DC Motor................................ Goals..............................................3

More information

Robust Heading Estimation Indoors

Robust Heading Estimation Indoors Technical report from Automatic Control at Linköpings universitet Robust Heading Estimation Indoors Jonas Callmer, David Törnqvist, Fredrik Gustafsson Division of Automatic Control E-mail: callmer@isy.liu.se,

More information

Robotics. Dynamics. Marc Toussaint U Stuttgart

Robotics. Dynamics. Marc Toussaint U Stuttgart Robotics Dynamics 1D point mass, damping & oscillation, PID, dynamics of mechanical systems, Euler-Lagrange equation, Newton-Euler recursion, general robot dynamics, joint space control, reference trajectory

More information

Estimation and Control of a Quadrotor Attitude

Estimation and Control of a Quadrotor Attitude Estimation and Control of a Quadrotor Attitude Bernardo Sousa Machado Henriques Mechanical Engineering Department, Instituto Superior Técnico, Lisboa, Portugal E-mail: henriquesbernardo@gmail.com Abstract

More information

Backstepping and Sliding-mode Techniques Applied to an Indoor Micro Quadrotor

Backstepping and Sliding-mode Techniques Applied to an Indoor Micro Quadrotor Proceedings of the 2005 IEEE International Conference on Robotics and Automation Barcelona, Spain, April 2005 Backstepping and Sliding-mode Techniques Applied to an Indoor Micro Quadrotor Samir Bouabdallah

More information

If the symmetry axes of a uniform symmetric body coincide with the coordinate axes, the products of inertia (Ixy etc.

If the symmetry axes of a uniform symmetric body coincide with the coordinate axes, the products of inertia (Ixy etc. Prof. O. B. Wright, Autumn 007 Mechanics Lecture 9 More on rigid bodies, coupled vibrations Principal axes of the inertia tensor If the symmetry axes of a uniform symmetric body coincide with the coordinate

More information

Design and modelling of an airship station holding controller for low cost satellite operations

Design and modelling of an airship station holding controller for low cost satellite operations AIAA Guidance, Navigation, and Control Conference and Exhibit 15-18 August 25, San Francisco, California AIAA 25-62 Design and modelling of an airship station holding controller for low cost satellite

More information

Nonlinear Identification of Backlash in Robot Transmissions

Nonlinear Identification of Backlash in Robot Transmissions Nonlinear Identification of Backlash in Robot Transmissions G. Hovland, S. Hanssen, S. Moberg, T. Brogårdh, S. Gunnarsson, M. Isaksson ABB Corporate Research, Control Systems Group, Switzerland ABB Automation

More information

Automated Tuning of the Nonlinear Complementary Filter for an Attitude Heading Reference Observer

Automated Tuning of the Nonlinear Complementary Filter for an Attitude Heading Reference Observer Automated Tuning of the Nonlinear Complementary Filter for an Attitude Heading Reference Observer Oscar De Silva, George K.I. Mann and Raymond G. Gosine Faculty of Engineering and Applied Sciences, Memorial

More information

CIS 4930/6930: Principles of Cyber-Physical Systems

CIS 4930/6930: Principles of Cyber-Physical Systems CIS 4930/6930: Principles of Cyber-Physical Systems Chapter 2: Continuous Dynamics Hao Zheng Department of Computer Science and Engineering University of South Florida H. Zheng (CSE USF) CIS 4930/6930:

More information

Single-track models of an A-double heavy vehicle combination

Single-track models of an A-double heavy vehicle combination Single-track models of an A-double heavy vehicle combination PETER NILSSON KRISTOFFER TAGESSON Department of Applied Mechanics Division of Vehicle Engineering and Autonomous Systems Vehicle Dynamics Group

More information

UAV Coordinate Frames and Rigid Body Dynamics

UAV Coordinate Frames and Rigid Body Dynamics Brigham Young University BYU ScholarsArchive All Faculty Publications 24-- UAV oordinate Frames and Rigid Body Dynamics Randal Beard beard@byu.edu Follow this and additional works at: https://scholarsarchive.byu.edu/facpub

More information

Autonomous Hovering with a Quadrotor Helicopter. Department of Electronic Systems. 8th Semester

Autonomous Hovering with a Quadrotor Helicopter. Department of Electronic Systems. 8th Semester Department of Electronic Systems 8th Semester Autonomous Hovering with a Quadrotor Helicopter Authors: Christian FinkPetersen Henrik Hansen Steen Larsson Lars Bo Theilgaard Madsen Michael Rimestad Supervisor:

More information

Quaternion based Extended Kalman Filter

Quaternion based Extended Kalman Filter Quaternion based Extended Kalman Filter, Sergio Montenegro About this lecture General introduction to rotations and quaternions. Introduction to Kalman Filter for Attitude Estimation How to implement and

More information

Multi-layer Flight Control Synthesis and Analysis of a Small-scale UAV Helicopter

Multi-layer Flight Control Synthesis and Analysis of a Small-scale UAV Helicopter Multi-layer Flight Control Synthesis and Analysis of a Small-scale UAV Helicopter Ali Karimoddini, Guowei Cai, Ben M. Chen, Hai Lin and Tong H. Lee Graduate School for Integrative Sciences and Engineering,

More information

Dynamics. 1 Copyright c 2015 Roderic Grupen

Dynamics. 1 Copyright c 2015 Roderic Grupen Dynamics The branch of physics that treats the action of force on bodies in motion or at rest; kinetics, kinematics, and statics, collectively. Websters dictionary Outline Conservation of Momentum Inertia

More information

Mathematical Modelling of Multirotor UAV

Mathematical Modelling of Multirotor UAV Mathematical Modelling of Multirotor UAV DENIS KOTARSKI, Mechanical Engineering, Karlovac University of Applied Sciences Trg J.J. Strossmayera 9, CROATIA, denis.kotarski@vuka.hr PETAR PILJEK, Faculty of

More information