Bayesian Tactile Exploration for Compliant Docking With Uncertain Shapes

Kris Hauser

Robotics: Science and Systems Conference, June 2018

Abstract. This paper presents a Bayesian approach for active tactile exploration of a planar shape in the presence of both localization and shape uncertainty. The goal is to dock the robot's end-effector against the shape -- reaching a point of contact that resists a desired load -- with as few probing actions as possible. The proposed method repeatedly performs inference, planning, and execution steps. Given a prior probability distribution over object shape and sensor readings from previously executed motions, the posterior distribution is inferred using a novel and efficient Hamiltonian Monte Carlo method. The optimal docking site is chosen to maximize docking probability, using a closed- form probabilistic simulation that accepts rigid and compliant motion models under Coulomb friction. Numerical experiments demonstrate that this method requires fewer exploration actions to dock than heuristics and information-gain strategies.

Download PDF pdf

Experiments on the physical RoboSimian.
Using the proposed tactile exploration controller, the RoboSimian robot attempts docking. Even with noise and occlusion, the first attempted site was successful. Purple: the sensed terrain shape as measured by an emulated laser sensor, with noise and localization error. Orange: the emulated force sensor output. Blue: the planned docking path.
In this example, the first attempted docking site was unsuccessful due to localization error and sensor noise, and slips. The second attempt is successful.
In this example, the first attempted docking site was unsuccessful by hitting underneath the hold. The second attempt is successful at choosing another hold.