Neural Networks for Robot Collision Estimation and Detection

This paper presents a mini-review on our previous work presented in ref. [1-6] in which the neural networks (NNs) were used for estimating and then detecting the robot’s collisions with the human operator during the cooperation task. This review investigates and compares the designed NN architectures, their application, the resulted mean squared error (MSE) from training, and their effectiveness (%) in detecting the robot’s collisions. This review reveals that the NN is an effective method in estimating and detecting the human-robot collisions