Adaptive locomotion in a complex environment: simulation of stick insect gap crossing behaviour

摘要

In a complex three dimensional environment, the ability to climb across large gaps and obstacles is fundamental for a stick insect. The same ability is desired for a walking animat that operates in a similar habitat, In this study, the neural network simulation WALKNET that has been modelled on stick insect walking behaviour is used as a basis for the introduction of new behaviours that enable it to climb over large gaps. Gap crossing behaviour of stick insects has been studied to show which behaviours are most crucial for reaching the far edge. As slowing down of forward movement and searching movements of the front legs in the gap have been considered most important, these two behaviours have been analysed. Subsequently, new modules for velocity control, adaptation of swing amplitudes and generation of searching movements have been implemented into the WALKNET controller. With these innovations, the animat is able to climb across gaps of more than twice its normal step length, using similar strategies as the biological model. The new behaviours improve its adaptability and performance under challenging environmental conditions. Further results of experimental gap crossing studies are discussed with respect to their value for the walking simulation and robotic applications.