Payload-Aware Trajectory Optimisation for Non-Holonomic Mobile Multi-Robot Manipulation With Tip-Over Avoidance
Florian Kennel-Maushart, Stelian Coros
Abstract
Cooperative mobile manipulation is an increasingly important topic in robotics: Just as humans need to collaborate on many tasks, robots need to be able to work together, e.g., to transport heavy or unwieldy objects in unstructured environments. But mobile multi-robot systems pose unique challenges, such as a much larger configuration space for motion planning, stability concerns and, especially for wheeled mobile robots, non-holonomic constraints. To tackle these challenges, we propose a multi-robot, bi-level optimization system, based on a direct transcription formulation for wheeled mobile manipulation. Our formulation uses static forces, calculated for stability objectives at the lower level to inform the wheeled trajectory planning at the higher level. This allows for effective planning, while ensuring safe execution, and improving the open-loop performance on real robots. We demonstrate our model's ability to tackle challenging motion-planning tasks and evaluate its improved real-world performance on the Clearpath Husky mobile platform. Finally, we integrate the system with our previously presented Mixed-Reality interface.