Datensatz

Zusammenfassung

Reinforcement learning is a natural choice for the learning of complex motor tasks by reward-related self-improvement. As the space of movements is high-dimensional and continuous, a policy parametrization is needed which can be used in this context. Traditional motor primitive approaches deal largely with open-loop policies which can only deal with small perturbations. In this paper, we present a new type of motor primitive policies which serve as closed-loop policies together with an appropriate learning algorithm. Our new motor primitives are an augmented version version of the dynamic systems motor primitives that incorporates perceptual coupling to external variables. We show that these motor primitives can perform complex tasks such a Ball-in-a-Cup or Kendama task even with large variances in the initial conditions where a human would hardly be able to learn this task. We initialize the open-loop policies by imitation learning and the perceptual coupling with a handcrafted solution. We first improve the open-loop policies and subsequently the perceptual coupling using a novel reinforcement learning method which is particularly well-suited for motor primitives.