Model-Free Learning Compensation of Robotic Arm Manoeuvres for a Free-Flying Base*

This study addresses the dynamic coupling between a servicer spacecraft and its robotic arm in which routine arm manoeuvres degrade the attitude performance of its base. The industrial robotic technique of iterative learning control is applied to base attitude control during repeated arm manoeuvres, compensating the corresponding disturbance torques without the need for complex dynamic models. Experiments carried out using Carleton University’s Spacecraft Proximity Operations Testbed demonstrate improved settling behaviour for the arm-base system as a whole, with base attitude control error reducing five-fold from 5.05° RMS in the first iteration to 0.81° RMS in the fourth iteration. Such improvements can reduce the keep-out volume needed by the servicer in spacecraft proximity operations, enabling transitions between standard arm configurations without endangering other spacecraft nearby.