Design and Low-Level control of a Humanoid Robot Using a Distributed Architecture Approach

This article describes methods and strategies used to develop a humanoid robot with a distributed architecture approach where centralized and local control co-exist and concur to provide robust full monitoring and efficient control of a complex system with 22 DOF. A description of the hardware is given before introducing the architecture, since that greatly influences the methods implemented for the control systems and helps in understanding the general decisions. The platform is still undergoing improvement, but the results are very promising, mainly because many potential approaches and research issues have presented themselves and will provide opportunities to test distributed control systems with possibilities that go far beyond the classical control of robots. Some practical issues of servomotor control are also considered since that turned out to be necessary before implementing higher levels of control-these are, in turn, addressed in the last part the article, which gives an example to demonstrate the possibility of keeping a humanoid robot in an upright balanced position using only local control after reaction forces on the ground. [PUBLICATION ABSTRACT]