Overview

This Chapter deals with navigation control from the perspective of analog computation. Two approaches are introduced: the first is based on Turing patterns and the second on autowaves. The two paradigms, Turing Patterns and autowaves, are solutions of a Reaction-Diffusion (RD) equation and are both implemented on a CNN. Therefore the first Section of the Chapter is devoted to a brief introduction to RD-CNNs.

The common idea underlying the two approaches is the use of a RD medium as an analog processor for robot navigation. The analog processor plays the role of a high-level control center or a pool of commands neurons, providing the robot with high-level capabilities for obstacle avoidance and trajectory planning.

The first example concerns the implementation of reactive behavior in an array of locally coupled command neurons through Turing patterns. The behavior of the robot is regulated by the status (the pattern) of the whole neuron ensemble rather than that of single neurons. In the second case, wave computing, an emerging paradigm of nonlinear science, is applied to the problem of robot navigation in a complex environment with obstacles.

As regards practical issues, the two approaches can be used to control the locomotion of the robots introduced in Chap. 5. However, for the sake of simplicity the experimental examples given in this Chapter deal with simple roving robots implemented using LEGO MindStorms which are gaining increasing interest in robotics as low-cost, easy-to-build and re-configurable mobile robot kits for educational and research objectives [Greenwald...