Performance Survey for Disturbances of a Controller using a Pulsed neural network for an Autonomous Mobile Robot

In prior research, we have shown that autonomous mobile robot controller using pulsed neural network (PNN) is able to achieve high performance than “sigmoid NNˮ which the output of neuron uses sigmoidal function, because PNN is able to consider the time axis. However, the controller using PNN is not able to satisfy performance yet. We expect that environmental disturbance such as the noise included in sensor values and slip of wheels are reduced the performance of the controller. Therefore we surveyed the performance for disturbances of the controller with an easy task. Then we verified that the performance of the controller using our proposed PNN (PNNSSP) is better than the performance of the controller using PNN. In this study, we survey the performance of the controller using PNN and PNNSSP by a difficult task. The results of computer experiments show that the performance of the controller using PNNSSP is better than the performance of the controller using PNN in the difficult task. However, PNN is too small a success rate of the task. Therefore, we need to survey more the performance of the controller using PNNSSP and the performance of the controller using PNN