Kinematic analysis of cat hindlimb stepping.

1. The purpose of this study is to analyze the kinematics of stepping from the point of view of limb control by attempting to dissociate the various neuronal and mechanical factors that shape actual movements observed experimentally. The cat hindlimb, like the primate arm, is a multijointed mechanical system with redundant degrees of freedom. We contend that a number of issues studied in the context of arm movement control, such as end point control and trajectory planning, may also be applicable to limb movement in locomotion. 2. We recorded and analyzed the kinematics of cat hindlimb movement in unrestrained over-ground locomotion at various speeds and with a variety of surface conditions. We found we could represent limb movement in the step cycle simply and concisely by the trajectories of the limb segment orientation angles. Each trajectory conformed to a monophasic sawtoothlike waveform, and the relative timing between segments was largely invariant in the range of movements studied. This contrasts with the representation using relative joint angles, as in Philippson's scheme, which exhibits a monophasic waveform at the hip but biphasic waveforms at the knee and ankle. 3. To investigate how whole limb kinematics, i.e., changes in the length and orientation of the whole limb, relates to that of the limb segments, we reconstructed limb movement from segment trajectories, assuming as a first approximation that they were indeed sawtooth wave forms. The result strongly suggested that the relative timing of segment movements played an important role in regulating limb length during the swing phase of the step cycle. A strong correlation between segment relative timing and changes in limb length was also observed experimentally. 4. A comparison between reconstructed and actual limb movement revealed two major differences. In contrast with the actual movement, which exhibited at least two extension phases and limb shortening during the stance phase, the reconstructed movement had only a single extension phase and no limb shortening. The discrepancies were fully accountable, however, by the limb-ground interaction in stance, indicating that the features present in the actual movement resulted from the limb-ground interaction rather than from any elaborated control by the nervous system. 5. A second difference between reconstructed and actual movements was evident in an apparent jerkiness of the former and a difference in the hindpaw paths during the swing phase. These differences could be accounted for by including the consequences of limb inertia and finite muscle power, namely a gradual rather than instant change in velocity. Using a bell-shaped velocity profile for the segment movements, we were able to accurately reconstruct limb kinematics during the swing phase. 6. We conclude from this analysis that the overall features of limb kinematics in stepping may be controlled by regulating a small set of parameters related to the orientation angles of the limb segments. Specifically, the position of the endpoint, the hindpaw in this case, may be determined by the relative timing and amplitudes of segment trajectories. The point-by-point details of the observed limb kinematics may be largely attributed to limb mechanics and the interaction of the limb with its environment. Thus the neural control may be simpler than the kinematics suggests.