Bioinspired Cognitive System with Multilevel Task/Subtasks Structure for Autonomous Navigation in Unknown Environments

The bioinspired cognitive scheme is a distributed processing architecture of three specialized blocks designed to emulate the basic function of the human brain, i.e, memory for generating and storing information about the environment and the state of the system, reason for solving problems in a model of the environment, and motion to execute the solution in the real world. In this paper, to extend the flexibility of this cognitive scheme applied to solve the problem of autonomous navigation in unknown environments with only distance sensors on board, we present an improved cognitive model that introduces the use of a stratified software architecture, where the layers, of decreasing complexity, describes different levels of functions required to displace a mobile robot in a self-discovered horizontal plane by recursive calls of the cognitive scheme. The software layers are organized in a pyramidal structure according to their priority after decompose the tasks in lower level sub tasks and up to basic actions. We introduce the proposed system, first, describing the basic cognitive scheme and then, the stratified software, explaining in particular, the functions designed to explore an unknown environment, to synchronize the real motion with respect to the virtual one, and to navigate in the discovered working space. We illustrate the performance of the system with experimental results obtained using an omnidirectional mobile robot in a horizontal plane.