During motion, the robotic snake can be caught and bent to a random shape. When released, it smoothly resumes motion.
Each limb controller has a miniature LED that is lit when that limb is applying force. This way it’s instantly obvious which parts of the trunk contribute the most to the motion .
Our goal is to distribute the computational complexity among the limbs. Rather than having “dumb” limbs manipulated by a fast central processor, we have “smart” limbs that do a substantial amount of calculations themselves. The experimental snake has 15 limbs, each equipped with a 25 MIPS processor and we try to develop algorithms that harness the parallelism of such a system.