Ankle, Hip and Stepping Strategies

A three-tier response to being pushed while balancing: move the ankle, swing the torso, or take a step.

These concepts come from research on human postural control. Horak and Nashner (1986) had subjects stand on a support surface that suddenly translated, and found that when standing normally, people mostly rotate the body about the ankle to recover balance — the ‘ankle strategy’; standing on a surface narrower than the foot, they switch to flexing mainly at the hip — the ‘hip strategy’; and with a large enough disturbance, people step to rebuild their base of support — the ‘stepping strategy.’ Humanoid push-recovery controllers borrow this same tiered scheme: the ankle strategy corresponds to adjusting the center of pressure (CoP) within the footprint; the hip strategy relies on rapidly rotating the upper body to generate angular momentum, equivalent to adjusting the centroidal moment pivot (CMP); and once the capture point (the ground point the robot must step to in order to stop completely) falls outside the support polygon, a step becomes unavoidable. Stephens (2007) derived analytic boundaries, based on these three strategies, for judging whether not stepping will necessarily lead to a fall.

ExampleA humanoid robot given a light push while standing can recover with ankle torque alone; a bigger push has it swing its torso or arms forward and back to absorb the impulse; a still bigger one forces a step in the direction of the push.

Also called

Ankle Strategy, Hip Strategy, Stepping Strategy

Related

Push Recovery · Balance Control · Capture Point · Center of Pressure (CoP) · Centroidal Moment Pivot · Support Polygon

Sources

Horak F. B., Nashner L. M. Central programming of postural movements: adaptation to altered support-surface configurations. J Neurophysiol, 1986
Stephens B. Humanoid push recovery. IEEE-RAS Humanoids, 2007
Push Recovery of a Position-Controlled Humanoid Robot Based on Capture Point Feedback Control (arXiv:1710.10598)