Cooperative Slip Detection Using a Dual-Arm Baxter Robot

When dealing with robotic manipulation tasks, slip detection and control is vital to overcome payload uncertainties and to compensate for external disturbances. Many modern smart manipulators are shipped with integrated joint torque sensing capabilities providing a potential means of detecting slip and hence generating a feedback signal for slip control, without the need for additional external sensors. This paper investigates preliminary results on slip detection with an aim to extend the existing work on single manipulator to a dual arm cooperative manipulator system. The slip signal is obtained through filtering of onboard joint torque measurements when different materials are held in a friction grip between the two cooperating robotic end-effectors of a Baxter robot.

Frequency domain analysis of data from preliminary experiments shows a substantial frequency component in the vicinity of 12 Hz when slip is taking place. Consequently, a 9–15 Hz band-pass Chebyshev filtered torque signal is proposed as a real-time slip detection signal and its performance evaluated in experiments using a selection of materials with differing mass, coefficient of friction and thickness.

Analysis of the resulting filter output and threshold signal for each material revealed that the proposed approach was effective at detecting slip for some materials (laminated card, MDF, PVC and Perspex) but not for others (sheet steel, cardboard and polypropylene). This is attributed to the variation in consistency in slip behaviour of different material/weight combinations.