An autonomous mobile agent by a chaotic neural network

We propose a network of chaotic neurons, whose dynamics is given by an external input, refractoriness and a temporal integration, and utilize as a motion generator of an autonomous mobile agent in an unpredictable irregular environment. An external input is given from a sensor to detect a surrounding environment. The network dynamics is shown to be chaotic in some parameter ranges, and enables a wider wandering compared with a random walk. An additional motion of targeting is implemented by an additional sensory input. Parameter dependence of the dynamics is examined, and a parameter is adaptively changed in various environments