Intelligent Autonomous Robotics: A Robot Soccer Case Study

===Artificial Intelligence相关链接===

Intelligent Autonomous Robotics: A Robot Soccer Case Study (Synthesis Lectures on Artificial Intelligence and Machine Learning)

ABSTRACT

Robotics technology has recently advanced to the point of being widely accessible for relatively low-budget research, as well as for graduate, undergraduate, and even secondary and primary school education. This lecture provides an example of how to productively use a cutting-edge advanced robotics platform for education and research by providing a detailed case study with the Sony AIBO robot, a vision-based legged robot. The case study used for this lecture is the UT Austin Villa RoboCup Four-Legged Team. This lecture describes both the development process and the technical details of its end result. The main contributions of this lecture are (i) a roadmap for new classes and research groups interested in intelligent autonomous robotics who are starting from scratch with a new robot, and (ii) documentation of the algorithms behind our own approach on the AIBOs with the goal of making them accessible for use on other vision-based and/or legged robot platforms.

KEYWORDS

Autonomous robots, Legged robots, Multi-Robot Systems, Educational robotics, Robot soccer, RoboCup

CONTENTS

1. Introduction

2. The Class

3. Initial Behaviors

4. Vision
4.1 Camera Settings
4.2 Color Segmentation
4.3 Region Building and Merging
4.4 Object Recognition with Bounding Boxes
4.5 Position and Bearing of Objects
4.6 Visual Opponent Modeling

5. Movement
5.1 Walking
5.1.1 Basics
5.1.2 Forward Kinematics
5.1.3 Inverse Kinematics
5.1.4 General Walking Structure
5.1.5 Omnidirectional Control
5.1.6 Tilting the Body Forward
5.1.7 Tuning the Parameters
5.1.8 Odometry Calibration
5.2 General Movement
5.2.1 Movement Module
5.2.2 Movement Interface
5.2.3 High-Level Control
5.3 Learning Movement Tasks
5.3.1 Forward Gait
5.3.2 Ball Acquisition

6. Fall Detection

7. Kicking
7.1 Creating the Critical Action
7.2 Integrating the Critical Action into the Walk

8. Localization
8.1 Background
8.1.1 Basic Monte Carlo Localization
8.1.2 MCL for Vision-Based Legged Robots
8.2 Enhancements to the Basic Approach
8.2.1 Landmark Histories
8.2.2 Distance-Based Updates
8.2.3 Extended Motion Model
8.3 Experimental Setup and Results
8.3.1 Simulator
8.3.2 Experimental Methodology
8.3.3 Test for Accuracy and Time
8.3.4 Test for Stability
8.3.5 Extended Motion Model
8.3.6 Recovery
8.4 Localization Summary

9. Communication
9.1 Initial Robot-to-Robot Communication
9.2 Message Types
9.3 Knowing Which Robots Are Communicating
9.4 Determining When A Teammate Is “Dead”
9.5 Practical Results

10. General Architecture
11. Global Map
11.1 Maintaining Location Data
11.2 Information from Teammates
11.3 Providing a High-Level Interface

12. Behaviors
12.1 Goal Scoring
12.1.1 Initial Solution
12.1.2 Incorporating Localization
12.1.3 A Finite State Machine
12.2 Goalie

13. Coordination
13.1 Dibs
13.1.1 Relevant Data
13.1.2 Thrashing
13.1.3 Stabilization
13.1.4 Taking the Average
13.1.5 Aging
13.1.6 Calling the Ball
13.1.7 Support Distance
13.1.8 Phasing out Dibs
13.2 Final Strategy
13.2.1 Roles
13.2.2 Supporter Behavior
13.2.3 Defender Behavior
13.2.4 Dynamic Role Assignment

14. Simulator
14.1 Basic Architecture
14.2 Server Messages
14.3 Sensor Model
14.4 Motion Model
14.5 Graphical Interface

15. UT Assist
15.1 General Architecture
15.2 Debugging Data
15.2.1 Visual Output
15.2.2 Localization Output
15.2.3 Miscellaneous Output
15.3 Vision Calibration

16. Conclusion

A. Heuristics for the Vision Module
A.1 Region Merging and Pruning Parameters
A.2 Tilt-Angle Test
A.3 Circle Method
A.4 Beacon Parameters
A.5 Goal Parameters
A.6 Ball Parameters
A.7 Opponent Detection Parameters
A.8 Opponent Blob Likelihood Calculation
A.9 Coordinate Transforms
A.9.1 Walking Parameters

B. Kicks
B.1 Initial Kick
B.2 Head Kick
B.3 Chest-Push Kick
B.4 Arms Together Kick
B.5 Fall-Forward Kick
B.6 Back Kick

C. TCP Gateway

D. Extension to World State in 2004

E. Simulator Message Grammar
E.1 Client Action Messages
E.2 Client Info Messages
E.3 Simulated Sensation Messages
E.4 Simulated Observation Messages

F. Competition Results
F.1 American Open 2003
F.2 RoboCup 2003
F.3 Challenge Events 2003
F.4 U.S. Open 2004
F.5 RoboCup 2004
F.6 U.S. Open 2005
F.7 RoboCup 2005

References
Biography
Password: ebooksclub.org
File size: 1.6 MB
Format: PDF

5