Predicting Force Redistributions on Walking Robots for Reliable Locomotion: Modeling and Experiments

A large number of walking robots walk with a statically-stable gait. A statically-stable walker has at least three feet that are in ground contact at any time. If there are more than three feet in ground contact, the normal (vertical) forces exerted by the ground on the feet of the walker are indeterminate, unless they are measured. Some walking robots may walk with more than three legs in ground contact in order to achieve greater stability. To ensure this stability it is desirable to predict how vertical forces passively redistribute underneath the feet during walker motions. Predictions of future foot forces can be used as a basis for accepting or rejecting any planned walker motion. Two methods — the least-squares method and the compliance method — for predicting this redistribution of forces in the face of static indeterminacy are presented in this work. Both methods are computationally efficient, and give reasonably accurate predictions, as verified by experiments on a walking robot.