Elegant stepping: A model of visually triggered gait adaptation

Existing visually guided walking machines have difficulty traversing terrai n cluttered with obstacles. These walking machines use computationally inte nse approaches that require construction of a geometrically correct model o f both the environment and the robot. However, most terrestrial vertebrates accomplish this task easily, suggesting that beater strategies exist. We p resent a model inspired by recent research in cats and humans. In our model , perception and action are tightly coupled. The mapping is adaptive and ba sed on experience. The goal of the adaptation is to use distance measuremen ts to smoothly modulate a central pattern generator (CPG) controlling gait. A key element in our model is the use of a temporal gating hypothesis. Thi s hypothesis simplifies the learning problem and is consistent with biologi cal observations. Our approach does not require that a geometric representa tion of the environment be created or updated based on new observations. Th is is in strong contrast to current practice in machine vision and robotics of surface reconstruction as a prerequisite to planning. Our simulation re sults indicate that the desired mapping can be learned quickly with Sew mis takes before perfect performance is achieved. The resulting gait modulation is smooth and coordinated with the phase of the CPG controlling the robot.