Locomotion in Complex Terrain

Abstract

Insects provide excellent models for agile legged locomotion. They use a six legged body plan to walk rapidly on virtually any surface. As they move through complex terrain the changes in direction and body attitude require alterations in the basic tripod gait described in earlier lectures. Our observations and those of several other workers suggests that these transitional behaviors require some interaction between local control circuits in the thoracic ganglia and commands that descend from the head ganglia. Turning requires asymmetrical changes in left and right legs that in turn generate asymmetrical ground reaction forces. In order to make these changes while the insect moves forward, some control from brain regions such as the central complex neuropiles may be required. This hypothesis is supported by findings that lesion in the central complex disrupt turning behaviors. To climb over blocks the animal must detect and measure the barrier, then alter leg movements to rear up and push its center of mass over the object. Bilateral lesions of the circumoesophageal connectives generate interesting deficits. On horizontal surfaces, these animals walk very normally. However, when challenged with blocks, they fail to make climbing movements prior to contacting the object. This problem can be readily explained by a loss of inputs from sensors on the head. However, problems with climbing up inclines are not as easily explained. The lesioned animals slip or fall over backwards on smooth 45 degree inclines and fail to bend their bodies appropriately when reaching the top of a block. Our observations suggest that these problems are due to difficulties in controlling body attitude. We have implemented many of these principles of insect locomotion in two lines of robots. The cockroach line employs legs with reasonably accurate segments and joints including the degrees of freedom that the animal typically uses for walking and climbing. These robots are powerful and have remarkable potential for agility, but their control architectures are complex and they are, therefore, still a work in progress. Whegs robots are simpler devices in which we abstract biological principles from insect locomotion. That is we take advantage of what the insect does, but do not constrain ourselves to do things in the exact same way. The appendages on these robots are compromises between wheels and legs. The three spoke design provides the rapid swing phase and slower stance of legs, but require only one drive motor for. the entire robot. Whegs robots are very fast and agile and control is much simpler. They can currently. perform a range of missions under radio control and can serve as a platform for research on more autonomous control strategies.