ADAPTIVE-CONTROL FOR BACKWARD QUADRUPEDAL WALKING .4. HINDLIMB KINETICS DURING STANCE AND SWING

1. Hindlimb step-cycle kinetics of forward (FWD) and backward (BWD) walking in adult cats were assessed. The hindlimb was modeled as a linked system of rigid bodies and inverse-dynamics techniques were used to calculate hip, knee, and ankle joint kinetics. For swing, net torque at each joint was divided into three components: gravitational, motion dependent, and a generalized muscle torque. For stance, vertical and horizontal components of the ground-reaction force applied at a point on the paw (center of pressure) were added to the torque calculations. Muscle torque profiles were matched to electromyograms (EMGs) recorded from hindlimb muscles. 2. Torque profiles for BWD swing were the approximate time reversal of those for FWD swing. At each joint, the net torque during swing was small because the mean motion-dependent and muscle torque components counteracted each other. At the hip a flexor muscle torque persisted except for a brief extensor muscle torque late in FWD swing and at the onset of BWD swing. At the knee the muscle torque was relatively negligible except for a peak flexor muscle torque late in FWD swing and early in BWD swing. At the ankle there was a midswing transition from a flexor to an extensor muscle torque during FWD swing and the reverse was true for BWD swing. 3. The vertical ground-reaction force was greater for the forelimbs than the hindlimbs during FWD stance; the reverse was true for BWD stance. Thus the hindlimbs bore a greater percentage (66%) of body weight than the forelimbs during BWD stance, and the forelimbs bore a greater percentage (59%) during FWD stance. For most of FWD stance, the hindlimb exerted a small propulsive ground-reaction force, but for BWD stance the hindlimb first exerted a braking force and then a propulsive force, with the transition occurring after midstance (59% of stance). 4. At the hip the ground-reaction force vector was oriented anteriorly and then posteriorly to the estimated joint center with a midstance transition during FWD stance. The muscle torque and joint power patterns showed similar transitions, changing from extensor and power generation to flexor and power absorption, respectively. For most of BWD stance the ground-reaction force vector was oriented anteriorly to the joint center and was counter-balanced by a large extensor muscle torque; nonetheless, power was absorbed because the hip flexed. 5. At the knee the ground-reaction force vector was oriented posteriorly to the estimated joint center for most of FWD stance and was counterbalanced by an extensor muscle torque that generated power as the knee extended. During BWD stance the ground-reaction force vector was oriented posteriorly and then anteriorly to the joint center with a midstance transition. The knee muscle torque transitioned at the same time from an extensor to a flexor torque and joint power from generation to absorption. At the ankle the ground-reaction force was oriented anteriorly to the joint center during most of FWD and BWD stance and was counter balanced by an extensor muscle torque combined with ankle extension to generate power. 6. With few exceptions, peak muscle torques (extensor vs. flexor) were associated with EMG activity of analogous muscles, and for both walking forms flexor muscle activity dominated the onset of swing and extensor muscle activity dominated stance. Differences in timing and amplitude that distinguish FWD-walking EMG patterns from those of BWD walking were consistent with the kinetic profiles. Details of the motor output may depend on unique interactions among supraspinal input, motion-related feedback, and a generalized motor pattern provided by a single spinal network for walking. Alternatively, a different network or specialized elements of the same network may be activated for each walking form.