Insights from Developing a Robotic Bimanual Manipulation System

Abstract

This thesis presents the development of a bi-manual robotic manipulation system utilizingopen-source robotics frameworks, including ros2 control and MoveIt2. In the realm of con- trol, we address key challenges such as coordinating multiple robots, independently manag- ing the hardware life-cycle via a state machine, and operating under limited communication bandwidth. For motion planning, we enhance sampling-based planners by incorporating goal set planning by leveraging the Task Space Regions (TSR) framework, enabling intuitive constraint representation. Additionally, we utilize IKFast, an analytical inverse kinematics solver, to compute inverse kinematics solutions in as little as one microsecond. Finally, we explore both synchronous and asynchronous execution strategies for trajectory execution in a dual-arm setup and discuss the scope for future work.