Tactile-Guided Automated Robotic Wiping

To automate the wiping task using robots, it is crucial to endow the robot with the capability to maintain desired contact with target surfaces. However, challenges arise when geometric priors and visual feedback are not available or when detailed object information is needed. In this paper, a tactile-guided hybrid pose/force control scheme based on gain-scheduling proportional-integral-derivative (PID) is proposed to address these challenges. Leveraging a tactile sensor, three features are extracted from the tactile data and used in the control loop. The control scheme regulates the contact normal force on unknown surfaces with slowly changing curvature while following a trajectory on the surface. It also allows real-time adaptation of the desired contact force to varying surface conditions. The performance of the proposed scheme is evaluated on a robot manipulator equipped with a tactile sensor. Experimental results show that the proposed system can identify stains, guide itself to re-clean areas, and successfully accomplish the wiping task.