Yuragi-based adaptive searching behavior in mobile robot: From bacterial chemotaxis to Levy walk

Abstract

Biologically inspired techniques have been widely investigated for realizing adaptive searching behaviour in a mobile robot. In this paper, we propose a framework based on ldquoYuragirdquo, a Japanese word for biological fluctuation. Yuragi-based framework is simple, has a noise-utilizing novelty to keep the robot searching for the goal, while its basic expression enables the robot to switch elegantly between stochastic and deterministic behaviour. It also doesn't need the model of the environment and therefore robust to changes in the environment. Here, we argue that the framework can realize effective searching behaviour, with and without sensory information. We do this by constructing a Yuragi-based equation that enables the robot to perform two already known effective searching behaviours: bacterial chemotaxis, probably the simplest yet effective gradient sources searching behaviour found in living creature, and Levy walk, specialized random walks with fractal movement trajectories that optimize random search for sparsely and randomly distributed target(s). As realized using the same framework, they can therefore be combined, and we will also show that it can be beneficial.