On generation of avoiding motion of robots which co-exist and cooperate with human (1st report, stochastic prediction of human motion and avoidance planning of manipulators

This paper proposes a control model for safety of human cooperative robots. Human motion is stochastically predicted by a Markov process model. Simulation results of the prediction corresponded to real human motion. Future dangerousness is estimated by using the predicted motion. Robot speed is changed in order to minimize the danger. Simulation results of the avoidance revealed that this control model is effective especially because prediction errors cannot cause extremely dangerous condition unlike by algorithms based on deterministic prediction.