Learning Structure, Reusability and Realtime Modeling in Teams of Autonomous Robots

1
Learning Structure, Reusability and Real-time
Modeling in Teams of Autonomous Robots

• New Work
• Transition to Segway RMP Platforms
• Transition to outdoor environments
• Single- and multi-agent learning in dynamic,
  uncertain and non-Markovian domains
• Fast, robust localization using MCL, Metropolis
  sampling, Energy-minimization techniques
• Adaptive fixed-team-plan selection for
  customization to unknown opponent

X - accomplished

• Achieved Goals
• CMVision Fast, color vision
• SRL Robust localization
• Plays Adaptive team strategy
• RRT Fast navigation planning
• WoLF Convergent multi-agent learning
• Multi-fidelity behavior
• Multiple open source code releases

X First RoboCup American Open
X RoboCup 2003
Improved sensor change segmentation
2003 2004
X Teleoperation of Segway RMP
Full Segway RMP team
Transfer single agent skills to Segway RMP
Transition to outdoor, vision based navigation