Robotic manipulation of massive objects in space

Abstract

This methodology is verified with air bearing table experiments that include object tracking, capture, path planning, trajectory generation, trajectory following, and berthing. Spatial berthing simulation results are also provided. Slip avoidance is also considered, and we contribute a methodology for slip avoidance via dynamic grasping, trajectory modification, and berthing force adjustment to avoid slip during multi-arm manipulation. Both planar air bearing and full spatial physical experiments are conducted to verify this methodology. We provide a more fundamental derivation of passivity-based stabilization methods that directly applies passivity theorem rather than relying on circuit-equivalent models. Furthermore, we contribute two additional methods for stabilizing hybrid simulations: H infinity control and adaptive control. Simulations are provided for a 1D scenario, and we investigate the benefits and drawbacks of each proposed method. Finally, we investigate the stability issue due to time delay in hybrid simulations.

Space tasks are attractive applications for autonomous and semi-autonomous robotic manipulation due to the risk and cost associated with astronaut resources. Payload restrictions lead to space manipulators exhibiting significant joint flexibility and limited motor actuation. This results in manipulation challenges in avoiding excessive oscillations during object transport, avoiding motor torque saturation (which can lead to trajectory following error and collision), and maintaining stable grasps with large inertial forces and torques due to payload motion. We investigate path and trajectory planning for flexible-joint manipulators. Our path planning approach is to utilize resolved velocity motion control and impose path curvature constraints to avoid oscillation during trajectory following and reduce required motor torque. We also contribute a model-based trajectory generation algorithm that treats the motor trajectory as the system input. Simulation results are used to show the efficacy the trajectory generation methodology, and physical experiments show how the feedforward motor trajectory can be used to reduce oscillations. We also contribute a methodology for outer-loop control that utilizes rigid-joint robot control techniques, but accounts for flexible-joint feasibility.