Planning and Control

Projects | | Links:

Adaptive Task Execution in Real-World Environments

Problem: Robustification of task execution by mobile manipulators in indudstrial environments
Challenge: Continuous change in environment configurations
Solution: Manipulability validation within dynamic 3D envionment and behavior tree-based recovery system

Developed a simulation module integrating Open3D-based dynamic environment generation with MoveIt! and RViz, enabling comprehensive manipulability analysis across 15 configurable parameters
Integrated a collision recovery algorithm and object 3D pose estimation-based action adaptation into a behavior tree for adaptive task execution
Set up a real-world demonstration framework by integrating hardware components, including Universal Robots UR5e with a mobile base, Robotiq-85 Gripper, and Azure Kinect Camera, with ROS2

Achieved a successful real-world demonstration of task adaptation to dynamic environmental changes
Achieved a successful real-world demonstration of task recovery from unexpected environmental interruptions