Sensorless collision detection based on friction model for a robot manipulator

Sang-Duck Lee,Jae-Bok Song

DOI: https://doi.org/10.1007/s12541-016-0002-3

IF: 2.0411

2016-01-01

International Journal of Precision Engineering and Manufacturing

Abstract:The importance of human-robot collision safety has been increasing recently due to the introduction of collaborative robots. However, conventional collision detection methods usually require additional sensors such as skin sensors, joint torque sensors, and acceleration sensors, which are impractical to implement due to their high cost. To address this problem, in this study we propose a collision detection method using only a manipulator’s encoder without any extra sensors. In the proposed scheme, the external torque due to collision is estimated using a generalized momentum-based observer and a friction torque model in the harmonic drive developed for a robot that is conducting position control. The performance of the proposed collision detection method was evaluated using a 6 DOF industrial manipulator. The experimental results show that collision can be reliably detected without any extra sensors for any type of robot manipulator.

engineering, mechanical, manufacturing