Attitude control in walking hexapod robots: an analogic spatio-temporal approach

In this paper, the control system of a biologically inspired walking robot is presented. An analog distributed system acts as Central Pattern Generator for the locomotion control, while the attitude control is performed by using a proportional integrative controller for each leg. The inclusion of a saturation block in the attitude control scheme allows to take into account the joint limits, leading to the formulation of a cellular non-linear network (CNN) based attitude controller. Thus, the whole control system can be structured as an analog control system realized by CNNs generating the locomotion pattern as a function of the sensorial stimuli from the environment. Both the network design and circuit realization are presented in this paper. Some experimental results obtained with the 18 degrees-of-freedom walking robot prototype are also reported. Moreover, some interesting experiments showing the robot capabilities to escape from unforeseen situations, in which overloading causes the saturation of a joint motor, are emphasized as an emerging result due to the action on both the locomotion control and the attitude control. Copyright (C) 2002 John Wiley Sons, Ltd.