Learning and imitation in heterogeneous robot groups / Wilhelm Richert. 2009
Content
1 Introduction
2 State of the art
2.1 Learning
2.2 Imitation
3 Architecture for learning and imitating in groups
3.1 Architectural overview
3.2 Layer interaction
3.3 Imitation in robot groups
3.4 Choice of the imitatee
3.5 Scenarios
4 Motivation layer
5 Strategy layer
5.1 Background
5.2 State of the art
5.3 Policy
5.4 State abstraction
5.5 Model
5.5.1 Transition heuristic
5.5.2 Failure heuristic
5.5.3 Reward heuristic
5.5.4 Simplification heuristic
5.5.5 Experience heuristic
5.6 Sample frequency
5.7 Exploration
5.8 Example
6 Skill layer
6.1 Two modes of operation
6.2 Component description
6.3 Configuration
6.4 Conclusion
7 An integrative example
7.1 Implementation of the motivation layer
7.2 Implementation of the strategy layer
7.3 Implementation of the skill layer
7.4 Evaluation
8 Imitation in robot groups
8.1 Related work
8.2 Overview of the multi-robot imitation approach
8.3 Transforming observations
8.4 Understanding observed behavior
8.5 Integrating recognized behavior
8.6 Evaluation
8.7 Conclusion
9 Choice of the imitatee
9.1 Related work
9.2 Background
9.3 Overview of the demonstrator choice process
9.4 Affordance detection
9.5 Affordance network generation
9.6 Comparing affordance networks
9.6.1 Structural difference of affordance networks
9.6.2 Parameter difference of affordance networks
9.6.3 Affordance network distance metric
9.7 Evaluation
9.7.1 Experimental setup
9.7.1.1 Parameterization of the environment
9.7.1.2 Affordances and their validation
9.7.1.3 Imitated behavior and how to measure its success
9.7.2 Selection experiment
9.7.3 Robustness experiment
9.7.4 Clustering experiment
9.8 Conclusion
10 Summary and outlook
A Notation
B Algorithms
List of Figures
List of Tables
Own publications
Bibliography