An Application of Reccurent Pulsed Neural Networks to Time-Series Processing of Autonomous Mobile Robot

In recent years, the study of time-series processing using Neural Networks has been attracted attention. Pulsed Neural Networks (PNNs) are suitable for time-series processing because they have integrator elements in themselves. Recurrent Neural Networks (RNNs) are also suitable for time-series processing because they have feedback loop in the networks. In this letter, we propose new Recurrent Pulsed Neural Networks (RPNNs) combining PNNs and RNNs in order to enhance time-series processing ability. We apply the proposed RPNNs to the controller of an autonomous mobile robot. Computer experiments indicate the efficacy of the proposed RPNNs