A System for Tracking and Locating Emergency Personnel Inside Buildings

Transcription

1 A System or Tracking and Locating Emergency Personnel Inside Buildings Ilir F. Progri, Student Member ION, William R. Michalson, Member ION, John Orr, and David Cyganski Electrical and Computer Engineering Department Worcester Polytechnic Institute 1 Institute Road Worcester, MA 169 Tel: (58) FAX: (58) iprogri@ece.wpi.edu or wrm@ece.wpi.edu Biography Mr. Ilir F. Progri received his Diploma o Engineer Degree in Electrical Engineering rom Polytechnic University o Tirana, Albania in 199. He received his MS degree in Electrical Engineering rom Electrical and Computer Engineering Department, Worcester Polytechnic Institute (WPI) in He is currently a graduate student o the Electrical and Computer Engineering Department, WPI, perusing his Ph.D. degree in Electrical Engineering with strong emphasis in carrier phase and signal processing, integrity monitoring, precision approach and landing, ilter design and implementation, numerical methods and numerical linear algebra, sotware development and management. Dr. William R Michalson is an Associate Proessor in the Electrical and Computer Engineering Department at Worcester Polytechnic Institute, where he also directs the Satellite Navigation Laboratory. The maority o his research ocuses on the development, test, and evaluation o GPS integrity monitoring algorithms, with an emphasis on integrity monitoring or solemeans navigation and precision approach. He is involved with the development o GPS systems or specialized applications. Previously, he was with Raytheon Company where he developed computer system architectures or space-based data and signal processors. Abstract One o the most hazardous obs ireighters and other emergency personnel must do is to enter a burning building. In this situation, the building loor plan may be unknown (and may change), visibility is impaired, and the locations o other equipment and personnel may change. Combining this situation with a limited air supply and potentially obstructed escape paths results in an extremely dangerous environment. This paper will discuss a system that is currently in development that uses GPS and GPS-like technologies to provide navigation inormation to emergency personnel inside a building and situation awareness monitoring to personnel outside a building. The paper will present the system architecture and any relevant simulated or experimental results. Introduction Recent advances in inormation technology and integrated electronics provide a means or locating ireighters and other emergency personnel inside buildings. In this paper, an approach to integrating the

2 technologies required to develop a location system are presented. From the ireighter s point o view this system must have characteristics such as: 1. Small size any equipment carried on the ireighter must be small enough to not interere with normal activity. Further, it must be rugged enough to survive extreme environments.. Flexibility the system must be useul in situations where there are no existing loor plans and where entry may occur in locations other than doors or windows 3. Communications position, health and situation awareness inormation may be communicated to, or rom, the ireighter.. Accuracy the accuracy o the system must be such that an individual may be located under zero visibility conditions. 5. Reliability the system must be reliable enough that ireighters are willing to entrust it with their saety. 6. Simplicity the system must be easy to use in a distracting environment. These characteristics orm a minimal set o criteria or a system designed or use in the extreme environment o a burning building. In such an environment there are strict limits on the amount o time a ireighter may be inside due to limitations imposed by a limited air supply and on increasing ire and smoke intensity. Additional system challenges result as a consequence o the complete or partial alling o the building, which may result in the destruction o escape paths. From the ireighter saety system point o view, a wearable device suitable or the environment must have the ollowing additional characteristics: 1. Identiy the current location (in three dimensions) to the incident command post outside the building under any circumstance.. Provide status (health and motion) inormation on each team member, and on the condition o the exit paths. 3. Provide emergency exit guidance (back-tracking) to each team member via synthesized voice commands, and to the incident command post. Provide "homing" signals to guide searchers in locating ireighters in trouble. Preliminary System Design Assume that the ireighter path(s) are those o igure 1 and igure. Figure 1: Fireighter scenario in a D area

3 Figure : Fireighter scenario in a 3D area Initially we assume that the itinerary (or path) o the ireighter inside the building is known with respect to a reerence coordinate system, which can be either chosen inside or outside the building. We will also assume that the ireighter is equipped with a wireless transceiver capable o tracking a GPS-like signal and transmitting inormation to each team member and to the incident command post. We will assume that pseudolites, or GPSlike signal emitters, can be utilized to accurately determine the location o the ireighter. We will also assume that path along which the ireighter is moving is either: 1. Visible by a single ixed receiver. Visible by multiple ixed receivers 3. Visible by no ixed receivers In the irst two cases, navigation system design can be based on the double dierence technique by using the measurements available rom a common set o visible pseudolites [1]. Similarly, double dierencing can be used in the second scenario case as well since it is reasonable to assume that the path has piecewise visibility to at least one ixed receiver located along the path. This technique is described in [1] in great detail. The last case can be treated utilizing the single dierence technique. Since the single dierence technique has not previously been applied in this application, we will provide additional theoretical background or this technique as an alternative navigating means. Single Dierences No ixed receivers The analytical ormulation o the carrier phase measurement between the th (ireighter) receiver and the th pseudolite, [ k], is as ollows, ( ) [ ] [ ] [ ] M [ k] = d [ k] + c $ t [ k] % $ t [ k] + # a + " k k. (1) + " The unknown quantities o expression (1) are deined in the ollowing order: d [ k] denotes the geometric distance between the th receiver and the th pseudolite. [ k] denotes the th receiver clock bias. t [ k] t similarly determines the th pseudolite clock bias. a [ ] represents the number o unknown cycles between the th receiver and the th pseudolite at the initial moment o tracking. M [ k] " depicts the multipath error (see expression 1 o [1]) o receiving the signal rom the th pseudolite rom paths dierent than the LOS path.

4 " [ k] denotes the carrier phase measurement error o the th receiver. k is the epoch index rom the moment o tracking the carrier phase o the th pseudolite. = c is also known as the wavelength o the carrier phase. A similar expression can be written or the carrier phase ormulation between the ireighter receiver th and the l th pseudolite as, M ( ) + # a [ ] + " [ k] [ k] l [ k] d [ k] + c $ t [ k] % $ t [ k] l l l l " = +. () Taking the dierence between the quantity (1) and () yields, % l % l % l M [ k] = d [ k] + c$ t [ k] + # a [ ] + " % l [ k] + " [ k] (3) % l. Assuming that or very short duration o l time the quantity c" t [ k] remains unchanged or changes very slowly then we can include this part in the ambiguity quantity as (known as modiied ambiguities) Thereore, assuming that the ilter states are the position and velocity o the ireighter receiver and modiied single dierence ambiguities we obtain the ollowing measurement (observable) vector/matrix equation, Ö [ k] H[ k][ s k] + e[ k]. (8) Where, H [ k] denotes the measurement matrix relating the state vector to the observable vector o the single dierence accumulated carrier phase. Assume that the system dynamics can be modeled as a Kalman ilter; thus, or a transition matrix T[k] the propagation state vector is ormulated with the help o the ollowing expression [], [ k +1 ] = T[ k][ s k] w[ k]. (9) s + The process noise w [ k] is assumed white sequence with known covariance Q[k] and uncorrelated with double carrier phase measurement noise, R[k]. Eliminating the intermediate steps, we present here the inal solution o the Kalman (recursive) ilter in two maor phases, l l l $ a [ ] # c" t [ k] + $ a [ ]. () Similarly, we can include the remainder o the multipath eect into the eect o noise [ k] "( ) [ k] " [ k] " # $ +. (5) M l IN Compute Residual Vector Compute Kalman Gain Vector Combining the result o () and (5) into (3) produces, $ l $ l l [ k] d [ k] # a $ = + [ ] + " [ k]. (6) Let M be the number o independent measurements. We can write (6) in vector/matrix notation as, Ö [ k] = d[ k] + a[ ] + e[ k]. (7) State Update Covariance Update Output Current Solution OUT Figure 3: Kalman ilter state update

5 State update (see igure 3) We seek to update the state in accordance with, ( ) [ k] = sˆ [ k] + G[ k] Ö[ k] H[ k][ sˆ k] sˆ IN Propagate the State Vector Propagate the Covariance [ k] [ k] z[ k] = s ˆ + G. (1) In equation (1), ŝ [ k] presents the state vector in the a priori estimate, z [ k] is deined as the residual vector; and the matrix G[k] denotes the Kalman gain matrix which in its optimal orm is written as, ( ) 1 T T [ k] = C[ k] H[ k] H[ k] C[ k] H[ k] + R[ k] (11) G. The update equations or the covariance matrix C [ k] in its sub-optimal and optimal orm are: [ k] = ( I G[ k] H[ k] ) C[ k] ( I G[ k] H[ k] ) T C Ready or Current Update upon the Arrival o New Measurement Data OUT Figure : Kalman ilter state propagation Simulation Description and Results Using the simulator developed in [1] a set o scenarios was developed or a ireighter moving in a building such as that depicted in igure 5. In this situation, the ireighter moves in dimensions. The movement was arbitrarily selected to move the ireighter irst to the east, then north, and then southwest. [ k] R[ k] G[ k] T + G, (1) [ k] = ( I G[ k] H[ k] ) C[ k] C. (13) Equations (1) through (13) orm the heart o the Kalman ilter state update phase. State update is ollowed by state propagation also known as time update or time propagation phase, Time update, state or time propagation (see igure ) One equation o the state or time propagation is obtained by modiying expression (9), RX Fixed Rx Thrown Pseudolites [ k 1] T[ k][ sˆ k] sˆ + =. (1) The state error covariance changes in accordance with, T [ k + ] = T[ k] C[ k] T[ k] Q[ k] C 1 +. (15) Figure 5: Simulated D tracking In the simulation, as the ireighter enters the building, a ixed receiver is placed at

6 near the point o entry. This establishes an arbitrary, but stationary, reerence point that will be used to establish a basis or mapping and or locating people and equipment inside the building. As the ireighter enters the building, additional pseudolite transmitters are literally scattered in arbitrary locations. These transmitters provide the additional signal sources needed or navigation. The transmitter positions need not be known a priori as they will be determined by the system (using the ireighter and the ixed receiver as a reerence). Scenario 1 moving east For this scenario we picture a ireighter who moves 3 m on the east direction as shown in igure 6. Fireighter traectory and pseudolite layout inside the building Y(M) Y(PL) Y(FIXED) X(M) Figure 6: Fireighter moving in the east direction The pseudolites are denoted with squares and the ixed receiver is presented with a circle. The ireighter moves in the east direction at 1cm per sec. The simulation results or this scenario are obtained by processing all the measurements that are obtained rom the receiver (pseudorange, carrier phase, and Doppler). lat pos err (m) PDOP PDO P vs sim time pdop 1 3 Figure 7: Fireighter PDOP Lateral position error dplat cplat cplat Figure 8: Fireighter lateral position error ver pos err(m) Vertical position error dpver cpver cpver Figure 9: Fireighter vertical position error

7 lat vel err(m/sec) Lateral velocity error dvlat cvlat cvlat Figure 1: Fireighter lateral velocity error ver vel err(m/sec) Vertical velocity error dvver cver cver Figure 11: Fireighter vertical velocity error Scenario moving north For this scenario we picture a ireighter who moves 6 m on the north direction as shown in igure 1. Fireighter traectory and pseudolite layout inside the building 5 Y(M) Y(PL) Y(FIXED) The pseudolites are denoted with squares and the ixed receiver is represented with a circle. The ireighter moves in the north direction at cm per sec. The simulation results or this scenario are obtained by processing all the measurements that are obtained rom the receiver (pseudorange, carrier phase, and Doppler). PDOP lat pos err (m) 3 1 PDO P vs sim time pdop 6 8 Figure 13: Fireighter PDOP Lateral position error.5 dplat cplat cplat Figure 1: Fireighter lateral position error Y(M) X(M) Figure 1: Fireighter moving in the north direction

8 . Vertical position error dpver cpver cpver For this scenario we picture a ireighter who moves about 13 m on the southwest direction as shown in igure 18. ver pos err(m) Figure 15: Fireighter vertical position error lat vel err(m/sec) Lateral velocity error. dvlat cvlat cvlat Figure 16: Fireighter lateral velocity error Y(M) Fireighter traectory and pseudolite layout inside the building Y(M) 8 Y(PL) Y(FIXED) X(M) Figure 18: Fireighter moving in the southwest direction The pseudolites are denoted with squares and the ixed receiver is represented with a circle. The ireighter moves in the southwest direction at 3 cm per sec. The simulation results or this scenario are obtained by processing all the measurements that are obtained rom the receiver (pseudorange, carrier phase, and Doppler). PDO P vs sim time 5 ver vel err(m/sec) Vertical velocity error..1 dvver cver cver PDOP 3 1 pdop Figure 18: Fireighter PDOP Figure 17: Fireighter vertical velocity error Scenario 3 moving southwest

9 lat pos err (m) Lateral position error dplat 1 cplat cplat Figure : Fireighter lateral position error ver pos err(m) Vertical position error. dpver cpver cpver Figure 1: Fireighter vertical position error lat vel err(m/sec) Lateral velocity error.6 dvlat cvlat cvlat ver vel err(m/sec) Vertical velocity error. dvver cver cver Figure 3: Fireighter vertical velocity error Conclusions and Future Considerations The work presented here represents a irst step towards realizing a system that is capable o providing the real-time location o emergency workers inside structures. The basic concepts o GPS are exploited in a way that allows 3D positioning without access to the GPS satellites. Several variations on the system presented are possible, and additional work to reine the overall system architecture is ongoing. Nevertheless, the simulations presented thus ar suggest that sub-meter location o workers is possible. Reerences 1. W. R. Michalson and I. F. Progri. "Assessing the Accuracy o Underground Positioning Using Pseudolites." Proceedings o the 13 th International Technical Meeting o the Satellite Division o the ION, ION GPS-, September 19-,, Salt Lake City, Utah.. R. G. Brown and P. Y. C. Hwang. Introduction to Random Signals and Applied Kalman Filtering. Copyright 1997, John Wiley & Sons, Inc Figure : Fireighter lateral velocity error