Optimal learning gain selection in model reference iterative learning control algorithms for human motor systems

The role of learning gains in the ability of a computational framework to better capture the behaviour of human motor control in learning and executing a task is the subject of discussion in this paper. In our previous work, a computational model for human motor learning of a task through repetition was established and its convergence analysed. In this paper, the performance of the model is investigated through the addition of degrees of freedom in selecting learning gains, specifically the ability to independently select the learning gain for the damping term. A particle swarm optimisation (PSO) algorithm is utilised to obtain a set of gains optimised to reduce the discrepancy between the experimental data and the simulated trajectories. It is found that it is possible to improve the accuracy of the computational model through the appropriate choice of learning gains. The results and interesting findings are presented and discussed in this paper.