Autonomous navigation of a mobile robot using a network of Hindmarsh-Rose (HR) neurons

In this paper, a hybrid control is proposed for autonomous navigation of a mobile robot in an unknown space with static and dynamic obstacles. This hybrid control is composed by the commutation between an avoidance control based on Hindmarsh-Rose (HR) neurons, which are interconnected by a diffusive coupling, and a posture tracking control. The commutation is in terms of the distance sensors that are incorporated on the frontal hemisphere of the mobile robot. The mobile robot control is implemented at wheel speeds level. A small coupling force is used to take advantage of the desynchronization on the membrane potentials when different inputs (external currents) are applied to each neuron. In this way, a greater sensitivity to any detection is obtained and a strategy of lateral avoidance is generated. On the other hand, a dynamic coupling is used to establish a strategy of frontal avoidance. Additionally, the performance of the proposed controller is numerically illustrated by means of computer simulations.