TY - JOUR
T1 - Hexapod Walking as Emergent Reaction to Externally Acting Forces
AU - El Sayed Auf, Adam
AU - Dudek, Nico
AU - Maehle, Erik
PY - 2009/1/1
Y1 - 2009/1/1
N2 - Insect-like walking with a six-legged robot in- volving 18 degrees of freedom is a fascinating as well as challenging task in terms of controlling and coordinating the 18 joints. Especially the control of those legs touching the ground, thus being mechanically coupled, is complex. Exter- nal forces acting to the walker's joints contain the information for each joint where and how far to move. This paper presents a decentralized controller approach measuring externally acting forces in each joint and combining active compliance with a step performing reflex. Local communication between the six legs as well as individual complying of the selected joints is used to achieve walking as an emergent reaction to the externally acting forces. This approach is based on an organic computing architecture and is implemented and tested on a six-legged walking machine. I. INTRODUCTION Observing insects moving with a surprising facility through extremely unstructured environments like branch- wood, leaves, and under branches can make a robotic engineer still jealous. Although great six-legged walking machines were built (1)(2), insects still outclass their robotic half siblings with their locomotion abilities. Insects are able to handle obstacles like large gaps or disturbances like losing foothold contact on slippery ground. They even compensate strong changes in their own body geometry like the loss of a leg. The fact that these capabilities in unstructured environments are still a challenge in robotic science leads to an increasing curiosity for the biological principles underlying those capabilities. For solving these challenges an internal accurate calculation by the insect's neurons seems hardly probable. Biological studies expose a decentralized reflex-based control system in strong com- bination with the organism's anatomical characteristics like elasticity and stiffness of muscles and joints. The later component is at least as important as the controller and has been disregarded for a long time. The deeper scientists probe into the biological coherences of animal locomotion the more important the anatomical features and qualities become. The muscles' elasticity seems to play an important part in terms of walking stability as well as compensating external forces acting on the walker's body. Especially the walking machine's legs touching the ground at the same time produce forces acting on the robot's body, which seem to be compensated in a biological organism. This compensation corresponds to the accurate calculation of a closed kinematic loop built up from the ground touching legs. Introducing a positive feedback concept (3)(4) in the walker's joints as successfully presented in two versions of the Walknet (5)(6) simulation may contribute to the improvement of walking robots. Positive feedback or active
AB - Insect-like walking with a six-legged robot in- volving 18 degrees of freedom is a fascinating as well as challenging task in terms of controlling and coordinating the 18 joints. Especially the control of those legs touching the ground, thus being mechanically coupled, is complex. Exter- nal forces acting to the walker's joints contain the information for each joint where and how far to move. This paper presents a decentralized controller approach measuring externally acting forces in each joint and combining active compliance with a step performing reflex. Local communication between the six legs as well as individual complying of the selected joints is used to achieve walking as an emergent reaction to the externally acting forces. This approach is based on an organic computing architecture and is implemented and tested on a six-legged walking machine. I. INTRODUCTION Observing insects moving with a surprising facility through extremely unstructured environments like branch- wood, leaves, and under branches can make a robotic engineer still jealous. Although great six-legged walking machines were built (1)(2), insects still outclass their robotic half siblings with their locomotion abilities. Insects are able to handle obstacles like large gaps or disturbances like losing foothold contact on slippery ground. They even compensate strong changes in their own body geometry like the loss of a leg. The fact that these capabilities in unstructured environments are still a challenge in robotic science leads to an increasing curiosity for the biological principles underlying those capabilities. For solving these challenges an internal accurate calculation by the insect's neurons seems hardly probable. Biological studies expose a decentralized reflex-based control system in strong com- bination with the organism's anatomical characteristics like elasticity and stiffness of muscles and joints. The later component is at least as important as the controller and has been disregarded for a long time. The deeper scientists probe into the biological coherences of animal locomotion the more important the anatomical features and qualities become. The muscles' elasticity seems to play an important part in terms of walking stability as well as compensating external forces acting on the walker's body. Especially the walking machine's legs touching the ground at the same time produce forces acting on the robot's body, which seem to be compensated in a biological organism. This compensation corresponds to the accurate calculation of a closed kinematic loop built up from the ground touching legs. Introducing a positive feedback concept (3)(4) in the walker's joints as successfully presented in two versions of the Walknet (5)(6) simulation may contribute to the improvement of walking robots. Positive feedback or active
UR - https://www.researchgate.net/publication/253528233_Hexapod_Walking_as_Emergent_Reaction_to_Externally_Acting_Forces
M3 - Journal articles
SP - 12
EP - 13
JO - Proceedings of Robotica - 9th Conference on Mobile Robots and Competitions
JF - Proceedings of Robotica - 9th Conference on Mobile Robots and Competitions
ER -