Home /Research /A universal algorithm for sensorless collision detection of robot actuator faults
MANIPULATION

A universal algorithm for sensorless collision detection of robot actuator faults

Saixuan Chen, Minzhou Luo, Feng He

Year
2018
Citations
30
Access
Open access

Abstract

When a robot is working properly, it is possible to collide with people or objects entering its working space. This research is different than usual control algorithm. It proposes a universal algorithm for sensorless collision detection of robot actuator faults to enhance the security of the robot. On the basis of the dynamic model, a classical friction model to ensure the accuracy of the whole dynamic model is introduced. This collision detection algorithm can conduct without any external sensors or acceleration and realize the real-time detection just needs to measure the motor current and the location information from the encoder of the robot joint. The value of external torque τ ext was used to compare with the threshold to detect the collision. After using the proposed collision detection method, the two rotational (2R) planar manipulators can detect the slight collision reliably. The experimental results and performance comparisons show that this sensorless collision detection algorithm is simple and effective. It can be promoted to any other type of robot arm with more degrees of freedom.

Keywords

CollisionCollision detectionRobotActuatorComputer scienceControl theory (sociology)AlgorithmEncoderAccelerationTorque

Related papers

Browse all MANIPULATION papers