CS 4630: Intelligent Robotics and Perception Case Study: Motor Schema-based Design Chapter 5 Tucker Balch
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology What We’ve Covered History of Intelligent Robotics (Chapter 1) Hierarchical paradigm (Chapter 2) Biological basis for behavior-based control (Chapter 3) Overview of behavior based control (Chapter 4) Subsumption architecture (Chapter 4) Motor schema-based control (Chapter 4)
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Upcoming Today: case study of behavior-based control for multirobot team. Friday: TeamBots tutorial, new project assignment Monday: Midterm Exam Weds: Begin Chapter 5 (Sensing) Friday: Guest Lecture (Koenig)
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Social Potentials Balch & Arkin, IEEE Transactions on Robotics and Automation, 2000
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology The Multi-Foraging Task
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Foraging Robots (1997) Balch, AI Magazine, Balch, Autonomous Robots, 2000.
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Foraging Robots (1997) Balch, AI Magazine, Balch, Autonomous Robots, 2000.
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Foraging Robots (1997) Balch, AI Magazine, Balch, Autonomous Robots, 2000.
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Foraging Robots (1997) Balch, AI Magazine, Balch, Autonomous Robots, 2000.
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Behavioral Sequencing SearchDeliver Red have red Aquire Red see red ~see red ~have red at red bin Acquire BlueDeliver Blue have blue see blue at blue bin ~see blue ~have blue
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Performance as Team Size Increases
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Problem: Inter-Robot Interference
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Heterogeneous Strategy 1: Specialization
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Heterogeneous Strategy 2: Territorial
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Performance Comparison
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Are Diversity and Performance Correlated? Need a measure of robot team diversity Approach: information theory
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Diversity and Performance Negatively Correlated in Foraging
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Diversity and Performance Positively Correlated in Soccer Homogeneous Team Heterogeneous Team
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Where We are Real-time Video Processing Behavioral Sequence Representation Learning Algorithms
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Observing and Modeling Live Multi- Agent Systems Motivation –Our agents should act intelligently in the presence of other agents: humans, external agents, adversaries Social insects: –Rich, multiagent interactions –Adversarial/territorial behaviors –Real biology in collaboration with entomologists
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Research Goal: Develop Algorithms That Enable Simultaneous tracking of all the individuals in a colony Recognition of individual and colony behaviors Learning of new single and multi-agent behavior models Application of the models to multi-agent software and robotic systems
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology The complexity of ant society Holldobler & Wilson, 1990 Gordon, 1999
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Video of ant behaviors
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Finding Ants In Images (1) CMVision: Color-based tracking –Initially developed for tracking soccer robots –Classifies and segments regions according to color –100s of regions, 32 colors, 30Hz, low cost Bruce, Balch & Veloso, IROS-2000
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Finding ants in images (2) Approach: background subtraction Enables classification by color and motion B ij = (1 - )B ij + I ij Background Image Current Image Movement
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Associating observations with individuals The association problem –Best optimal algorithm O(n 3 ) –Greedy approach O(n 2 ) Noisy data presents additional challenges –Splitting, merging, drop-outs, pop-ups Current approach –“Greedy agents” leverage domain knowledge Future –Parallel implementations, Bayesian techniques (e.g. Xiang & Lesser), radar tracking techniques
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Analyzing the Spatial Behavior of a Colony
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology No Food Available
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Food Available
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Vector Representation
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Recognition Task: Right Turn, Left Turn, Straight? Approach: Average turning angles over a window Classify turns according to average: –if A < - , right turn –if A > , left turn –otherwise, straight n is the window size
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Example
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Recognizing Behavior from Movement Traces Hypothesis: –Observed movement features considered over time can be used to classify the behavior of a physical agent Previous success in observation of soccer agents –Hidden Markov Models (Han & Veloso, 1999) Example features –binary: towards-food, at-food, towards-home, at- home –continuous: velocity, turn-rate, path randomness Example behaviors –foraging, patrolling, carrying, recruiting
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Hidden Markov Model Representation S1S3 S A B C AAABBBBBBBBBBBBCCABBBBBBBBBCCA
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Hidden Markov Model Representation S1S3 S A 0.8 B 0.1 C 0.1 A 0.1 B 0.8 C 0.1 A 0.1 B 0.1 C 0.8 ACABBBABBBCBBBBCAABBBBABBBBCCA
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Using HMMs for Recognition With the Viterbi Algorithm AAABBBBBBBBBBBBCCABBBBBBBBBCCA
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology inspired by Han & Veloso, 1999
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Real-time Video Processing Behavioral Sequence Representation Recognition Algorithms Learning Algorithms
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology Thanks to Zia Khan Manuela Veloso James Bruce Gak Kaminka Pat Riley Rande Shern Ashley Stroupe DARPA Control of Agent Based Systems (CoABS)
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology
Observing Ants: Tracking and Analyzing the Behavior of Live Insects Tucker Balch Collaborative Perception and Robotics Lab
AMiRE October, 2001 Tucker Balch Georgia Institute of Technology