Improving the performance of compliant motions by online geometric uncertainty reduction

The automatic planning of robotic assembly tasks usually gives rise to a sequence of compliant motions. These motions are executed using online sensory information to lessen the effect of modelling and sensing uncertainties affecting the task. The paper presents a procedure to estimate some geometric parameters in a planar assembly task in order to improve the performance of compliant motions. This procedure is part of a two-phase fine-motion planner for robotic assembly tasks in the plane, developed by the authors.