Optimised Learning from Demonstrations for Collaborative Robots

The approach of Learning from Demonstrations (LfD) can support human operators especially those without much programming experience to control a collaborative robot (cobot) in an intuitive and convenient means. Gaussian Mixture Model Regression (GMM GMR) are useful tools for implementing such LfD approach. However, well-performed GMM/GMR require series demonstrations trembling jerky features, which challenging achieve actual environments. To address this issue, paper presents novel optimised improve clusters then further so that enabled cobots carry out variety complex manufacturing tasks effectively. This research has three distinguishing innovative characteristics: 1) noise strategy is designed scatter with features better the optimisation GMM/GMR; 2) Simulated Annealing-Reinforcement (SA-RL) based algorithm developed refine number eliminating potential under-/over-fitting issues on 3) B-spline cut-in integrated GMR adaptability reproduced solutions dynamic tasks. verify approach, cases studies pick-and-place different complexities were conducted. Experimental results comparative analyses showed exhibited good performances terms computational efficiency, solution quality adaptability.