Distributed Planning and Execution for Flexible Robotic Assembly

Institute for Complex Engineered Systems
Carnegie Mellon University

 Traditional planning and scheduling approaches do not work well for flexible robotic assembly environments. These approaches are  usually pre-planned and cannot adapt to the dynamic changes of the system. In this project, we adopt an agent based approach to address the issues on fault-tolerance, uncertainty-handling, interleaving of planning and execution, easy-to-implement and easy-to-maintain, etc. We have developed an any-time algorithm based on the A* search algorithm to reduce the idle time of each resource (robot). The agent-based system uses Task Control Architecture (TCA) developed at Carnegie Mellon University as message-passing mechanism and uses RobotWorld as our testbed.

System Architecture:


System Architecture

Planning Result for a 20-operation task:


Planning Result

Execution Result for the same task:


Execution Result