MEAM 510 Fall 2011 Bruce D. Kothmann

Similar documents
MEAM 510 Fall 2012 Bruce D. Kothmann

Inverted Pendulum. Objectives

Acceleration Feedback

General procedure for formulation of robot dynamics STEP 1 STEP 3. Module 9 : Robot Dynamics & controls

Mechatronic System Case Study: Rotary Inverted Pendulum Dynamic System Investigation

DC-motor PID control

University of Utah Electrical & Computer Engineering Department ECE 3510 Lab 9 Inverted Pendulum

THE REACTION WHEEL PENDULUM

Review: control, feedback, etc. Today s topic: state-space models of systems; linearization

Lecture 6: Control Problems and Solutions. CS 344R: Robotics Benjamin Kuipers

Application Note #3413

Rotary Inverted Pendulum

Harmonic Oscillator. Mass-Spring Oscillator Resonance The Pendulum. Physics 109 Experiment Number 12

Feedback Control of Linear SISO systems. Process Dynamics and Control

MECHATRONICS ENGINEERING TECHNOLOGY. Modeling a Servo Motor System

The loop shaping paradigm. Lecture 7. Loop analysis of feedback systems (2) Essential specifications (2)

(Refer Slide Time: 00:01:30 min)

H :50 TR Hall H12 2-3:50 TR Cronk H :50 TR Swanbom H14 8-9:50 TR Barker :50 MW Swanbom :50 TR Cronk :50 TR Nelson

Angle estimation using gyros and accelerometers

Digital Pendulum Control Experiments

Multi Rotor Scalability

Automatic Control Systems. -Lecture Note 15-

Reglerteknik, TNG028. Lecture 1. Anna Lombardi

Angle estimation using gyros and accelerometers

Robotics. Dynamics. Marc Toussaint U Stuttgart

Chapter 7 Control. Part Classical Control. Mobile Robotics - Prof Alonzo Kelly, CMU RI

Example: DC Motor Speed Modeling

Project Lab Report. Michael Hall. Hao Zhu. Neil Nevgi. Station 6. Ta: Yan Cui

DC Motor Position: System Modeling

System simulation using Matlab, state plane plots

EE 474 Lab Part 2: Open-Loop and Closed-Loop Control (Velocity Servo)

The basic principle to be used in mechanical systems to derive a mathematical model is Newton s law,

Feedback Basics. David M. Auslander Mechanical Engineering University of California at Berkeley. copyright 1998, D.M. Auslander

State Space Representation

FUZZY LOGIC CONTROL Vs. CONVENTIONAL PID CONTROL OF AN INVERTED PENDULUM ROBOT

(Refer Slide Time: 1:42)

Department of Mechanical Engineering

Linear control of inverted pendulum

Today. Why idealized? Idealized physical models of robotic vehicles. Noise. Idealized physical models of robotic vehicles

Chapter 12 Static Equilibrium

α = p = m v L = I ω Review: Torque Physics 201, Lecture 21 Review: Rotational Dynamics a = Στ = I α

Torque. Introduction. Torque. PHY torque - J. Hedberg

SRV02-Series Rotary Experiment # 7. Rotary Inverted Pendulum. Student Handout

2.004 Dynamics and Control II Spring 2008

Embedded Control: Applications and Theory

Real-Time Implementation of a LQR-Based Controller for the Stabilization of a Double Inverted Pendulum

Example: Modeling DC Motor Position Physical Setup System Equations Design Requirements MATLAB Representation and Open-Loop Response

Control Systems I. Lecture 2: Modeling. Suggested Readings: Åström & Murray Ch. 2-3, Guzzella Ch Emilio Frazzoli

Lab 6d: Self-Erecting Inverted Pendulum (SEIP)

Robot Manipulator Control. Hesheng Wang Dept. of Automation

FEEDBACK CONTROL SYSTEMS

Laboratory Exercise 1 DC servo

A consideration on position of center of ground reaction force in upright posture

ELEC4631 s Lecture 2: Dynamic Control Systems 7 March Overview of dynamic control systems

MAE143a: Signals & Systems (& Control) Final Exam (2011) solutions

EE Homework 3 Due Date: 03 / 30 / Spring 2015

Robotics. Dynamics. University of Stuttgart Winter 2018/19

Physics 201, Lecture 21

CS491/691: Introduction to Aerial Robotics

Solved Problems. Electric Circuits & Components. 1-1 Write the KVL equation for the circuit shown.

Harmonic Oscillator. Outline. Oscillatory Motion or Simple Harmonic Motion. Oscillatory Motion or Simple Harmonic Motion

Unit 8: Part 2: PD, PID, and Feedback Compensation

Overview of Experiments for Magnetic Torque

Lecture «Robot Dynamics»: Dynamics and Control

Classical Dual-Inverted-Pendulum Control

Balancing of an Inverted Pendulum with a SCARA Robot

Physics 100 Reminder: for on-line lectures

ECEN 420 LINEAR CONTROL SYSTEMS. Lecture 6 Mathematical Representation of Physical Systems II 1/67

Simple Harmonic Motion

Introduction to centralized control

Physics 101 Lecture 12 Equilibrium

Lab 3: Quanser Hardware and Proportional Control

Satellite Attitude Control System Design Using Reaction Wheels Bhanu Gouda Brian Fast Dan Simon

Lecture 12. Upcoming labs: Final Exam on 12/21/2015 (Monday)10:30-12:30

Lab 6a: Pole Placement for the Inverted Pendulum

Announcements Oct 16, 2014

Appendix A: Exercise Problems on Classical Feedback Control Theory (Chaps. 1 and 2)

Contents. PART I METHODS AND CONCEPTS 2. Transfer Function Approach Frequency Domain Representations... 42

Rotational Kinetic Energy

Unit 1: Equilibrium and Center of Mass

Double Inverted Pendulum (DBIP)

Designing Information Devices and Systems II Spring 2018 J. Roychowdhury and M. Maharbiz Homework 7

MinSeg balancing using pole-placement

Lecture 1: Introduction to System Modeling and Control. Introduction Basic Definitions Different Model Types System Identification

Modelling and Control of DWR 1.0 A Two Wheeled Mobile Robot

Simple and Physical Pendulums Challenge Problem Solutions

Welcome back to Physics 211

Lecture «Robot Dynamics»: Dynamics 2

Torques and Static Equilibrium

D(s) G(s) A control system design definition

Introduction to centralized control

Control of Mobile Robots

Chapter 6. Applications of Newton s Laws

Unit 7: Oscillations

Oscillations. Phys101 Lectures 28, 29. Key points: Simple Harmonic Motion (SHM) SHM Related to Uniform Circular Motion The Simple Pendulum

Chapter 15 Oscillations

Application of Neural Networks for Control of Inverted Pendulum

H-infinity Model Reference Controller Design for Magnetic Levitation System

6.003: Signals and Systems

Review for 3 rd Midterm

Transcription:

Balancing g Robot Control MEAM 510 Fall 2011 Bruce D. Kothmann

Agenda Bruce s Controls Resume Simple Mechanics (Statics & Dynamics) of the Balancing Robot Basic Ideas About Feedback & Stability Effects of Proportional Feedback Two Key Observations About Integral Feedback Derivative Feedback Used to Stabilize Why PID Feedback of Arm Angle Won t Work Effect of Sensor Alignment Effect of Sloped Table What Will Work Some Implementation Issues Hall Effect Sensor for Wheel Angle Dealing with Accelerometer Noise & Rate Gyro Bias 10-Bit A/D and Amplifiers Sample Rate : Is Faster Always Better? Best Robots Invited to Demo in ESE 406 This Spring! MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 1

Inverted Pendulum MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 2

Inverted Pendulum Description M1 Microcontroller (PID Control) Encoder From Lego Motor 43362 EncoderGeek.com http://www.philohome.com/motors/motorcomp.htm MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 3

Ping Pong Poise MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 4

Ping Pong Poise Description This is a Fixed-Point Regulator Servo Motor Optical Voltage Divider M1 Digital Microcontroller (PD Control) MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 5

My Other Controls Design Experience MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 6

Simple Mechanics Part 1 : Static Equilibrium Q=Motor Torque Pin Shear Force 2r Weight F= Friction Normal Force l FBD of Body Weight Pin Shear Force Q=Motor Torque Sum Moments @ Center of Wheel Q rf Sum Moments @ Pin in Body Q mglsin mgl Sum Forces Parallel to Slope on Both Objects (Pin Shear Drops Out) sin M r F Mg Mg ml MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 7

Simple Mechanics Part 2 : Dynamic Instability l Weight Technically WRONG 2 but Qualitatively Sort of Close to Right Dynamics of the Body dt Q=Motor Torque d I Qmglsin Qmgl 2 Main Problem : Gravity is Like a Negative Spring If Pendulum Starts Falling, Gravity Makes it Fall Faster (Control Lawyers Call This a Dynamics Problem, Not Disturbance Problem) MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 8

Effects of Proportional Feedback 2 d Q K P I K 2 P mgl dt Proportional Feedback Gain Has To Be Large Enough to Create Net Positive Spring (Overcome Open-Loop Negative Spring) Beware of Destabilizing Effect of Delay (Due to Processing Time or Other Dynamics) Qt ( T) K ( t) P Position @ t+t Position @ t MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 9

Two Key Observations About Integral Feedback To Reach Steady State (Equilibrium), Input to Integral Feedback MUST Be Zero! This Is Why Integral Feedback is so Powerful! This Is Why Integral Feedback is so Peril-ful! Same Speed Requires More Throttle Example : Cruise Control Encountering a Hill Integral Feedback Throttle Angle Change Proportional Feedback Speed Change Integral Feedback is Very Often Destabilizing (Including in This Problem!) MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 10

Effect of Sensor Alignment Attitude Sensor Won t Be Perfectly Aligned with Line Between Hinge & CG Static Equilibrim (See Below) Cannot Achieve =0! Even Integral Feedback (Which Acts Like K P Infinity) Fails! Cart ALWAYS Slowly Drifts Away! Q K K mgl P M P K mgl K P P M Effect Exaggerated for Clarity M MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 11

Effect of Slope M r ml Static Equilibrium (Overall CG Must Be Above Wheel Contact Point) Don t Want =0! Integral Feedback of Cannot Work on a Slope! Could Achieve Static Equilibrium with Exactly Right Value of Proportional Gain Only, But This Would Not Be Enough for Stability! Q Mgr mgl K P Torque Due to Feedback Must Equal Torque Required to Stay Upright MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 12

Human Operator! What Might Work? Use Wireless To Allow Operator to Command Lean Angle On Level Ground, Operator Cancels Installation Misalignment On Tilt, Operator Commands Lean Into the Hill Sufficient to Use PD Only? System Not Stable Without Operator, But Instability Slow Enough to Be Easily Compensated (Many Airplanes & Bicycles Are Like This) Use Wheel-Angle Feedback! Q K K P W Probably bl Want Derivative of Wheel Angle Too for Stability Arbitrary Steady Torque Can Be Achieved by Small Position Drift (Wheel Angle Change From Power-On Condition) This Fixes Both Installation Misalignment & Slope Effects! If Gains Make System Stable (You Need Equations to Know for Sure), Required Position Drift Happens Automagically! Hall Effect Sensor Can Measure Something Like (); Maybe Okay to Use PD Feedback on This? MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 13

Measuring Body Angle Accelerometer Sees Gravity as Negative Acceleration (General Relativity) But Accelerometer Also Sees Local Linear Acceleration (Due to Wheel Movement & Angular Acceleration Times Moment Arm) Accelerometer Has Noise, Especially at High Frequency 2 d a awheel r gsin 2 dt Body Angle Can Also Be Estimated by Integrating Angular Velocity, But This Will Drift (Because Angular Velocity Won t Read Exactly 0 at Rest) t (0) ( ) d 0 a MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 14

Complementary Filter : Combine a & Basic Idea Use Integral of angular velocity at high frequency (where a is noisy) Use (-a/g) at low frequency (where drift of angular velocity is bad) Theta = High Pass Filter Of + Integral of Low Pass Filter Of (-a/g) Need Digital Implementation of Integral & Filters Integrate High-Pass i i1 it ˆ ˆ i i1 i i1 Combine ˆ ˆ i i i Low-Pass ˆ ˆ 1 / i i 1 a i g MEAM 510 : Balancing Robot Control BDK : 2011-10-26 Page 15

Miscellaneous Implementation Issues Scaling of accelerometer voltage into A/D: you really only care about +/- ~5 deg, so make sure that range is 0 to 5 volts so you make good use of your 10-bit A/D (1024 Values) build a simple op-amp circuit? Very high processor speeds may cause problems, because digital filters may require very high precision. Also, digital filters are easiest to design with a fixed frame rate. MATLAB has a very convenient C2D function for converting analog filters to digital. Do you need to go faster than ~1000 Hz? I told a couple of lies earlier motor friction might allow proportional feedback to work on level ground & carpet might allow equilibrium on small slopes! But these effects are very unreliable (low robustness) MechaMiniSegway Notes BDK : 26-Oct-2009 : Page 16