Tutorial on Embodiment
220.127.116.11. Leg coordination in insect walking*
Leg movements in insects are controlled by largely independent local neural circuits that are connected to their neighbors, there is no central controller that coordinates the legs during walking. The leg coordination comes about by the exploitation of the interaction with the environment (Cruse, 1990; Cruse et al., 2002). If the insect stands on the ground and moves forward by pushing backwards with one of its legs, as an unavoidable implication of being embodied, all the joint angles of the legs standing on the ground will instantaneously change. The insect's body is pushed forward, and consequently the other legs are also pulled forward and the joints will be bent or stretched. This fact can be exploited to the animal's advantage. All that is needed is angle sensors in the joints - and they do exist - for measuring the change, and there is global communication between the legs! But the communication is through the interaction of the agent with the environment, not through neural processing.
Inspired by the fact that the local neural leg controllers need only exploit this global communication, a neural network architecture called WalkNet has been developed which is capable of controlling a six-legged robot (Dur et al., 2003) (Video 18.104.22.168.1 and 22.214.171.124.2). This instance of morphological computation takes over part of the task that would have to be done by the brain - the communication between the legs and the calculation of the angles on all the joints - is performed by the interaction between the insect and the world.
Video 126.96.36.199.1. A kinematic simulation of an insect controlled by Walknet. (courtesy Holk Cruse)
Video 188.8.131.52.2. A kinematic simulation of an insect controlled by Walknet when walking over obstacles. (courtesy Holk Cruse)
*This case study has previously appeared in Pfeifer & Gomez, 2009.
Cruse, H. (1990). What mechanisms coordinate leg movement in walking arthropods?. Trends in Neurosciences 13, 15-21.
Cruse, H., Dean, J., Durr, V., Kindermann, T., Schmitz, J. & Schumm, M. (2002). Neurotechnology for biomimetic robots, MIT Press, chapter A decentralized, biologically based network for autonomous control of (hexapod) walking, pp. 384-400.
Dur, V., Krause, A. F., Schitz, J. & Cruse, H. (2003). Neuroethological concepts and their transfer to walking machines. Int. J. Robotics Research 22, 151-67.
Pfeifer, R. & Gomez, G. (2009). Morphological computation - connecting brain, body, and environment. In B. Sendhoff, O. Sporns, E. Körner, H. Ritter, & K. Doya, K. (eds.), Creating Brain-like Intelligence: From Basic Principles to Complex Intelligent Systems (pp.66-83). Berlin: Springer.