We empirically investigate the implications of applying Raibert’s Scissor algorithm to the Spring Loaded Inverted Pendulum (SLIP) model in combination with other controllers to achieve transitional maneuvers. Specifically, we are interested in how the conjectured neutral stability of Raibert’s algorithm allows combined controllers to push the system’s operating point around the state space without needing to expend limited control affordance in overcoming its stability or compensating for its instability. We demonstrate 2 cases where this facilitates the construction of interesting transitional controllers on a physical robot. In the first we use the motors in stance to maximize the rate of change of the body energy; in the second we take advantage of the local environmental energy landscape to push the robot’s operating point to a higher or lower energy level without expending valuable motor affordance. We present data bearing on the energetic performance of these approaches in executing an accelerate-and-leap maneuver on a monopedal hopping robot affixed to a boom in comparison to the cost of anchoring the robot to the SLIP template.
For more information: Kod*lab