Dynamic Balance

Dynamic balance control monitors the robot's center of mass velocity and momentum, not just position. Unlike static balance (where the center of mass stays above support), dynamic balance allows the center of mass to move outside support temporarily - like humans do during running. The controller uses predictive models to anticipate where the robot will be, not just where it is. Reaction forces from ground contact are measured and used to compute balance adjustments. Strategies include ankle torque modulation (small perturbations), hip strategy (larger shifts), and stepping (placing new support point). The inertial measurement unit (IMU) provides high-rate orientation feedback enabling rapid corrective actions.

Dynamic balance enables humanoid robots to recover from pushes without falling. Running gaits require dynamic balance as the center of mass periodically leaves the support polygon. Walking on moving platforms (ships, trains) adapts to external accelerations. Rapid direction changes during navigation use momentum control. Catching thrown objects involves dynamic balance during reaching. Jumping and landing require prediction and rapid stabilization. Manipulation of heavy objects uses dynamic balance to counteract shifting loads. Sports and entertainment robots perform acrobatic movements through dynamic balance mastery.