Bio-inspired control of legged locomotion enhanced using minimal feedback (Technion)
Various methods in developing legged robots that are able to walk on a variable environment have been developed in the past years, including quasi-static locomotion; dynamic locomotion with feed-forward control; feedback control; or with combined feed-forward and feedback control. However, all of the above approaches carry limitations. All the current solutions operate exclusively on horizontal planes and are relatively slow and non-human like. In order to be robust, these methods require large computation resources, as well as accurate sensors and actuators.Our method uses coupled oscillators with discrete output to generate patterns of events which define the time of activation of the robot's motors. This produces a stable, energy efficient walking gait, taking advantage of the system's inertia. The strength and duration of activation of the motors are modified based on minimal feedback describing the surface's slope. This allows the robot to walk on a wide range of slopes. Current control of a knee-less biped model is capable of navigating -9° to +9° slopes, similar to most human environments. A wider range of slopes and other gaits, including climbing, are expected for a kneed biped model. The biological likeness of the design makes it useful in patient rehabilitation. The proposed control method can be used to improve the robustness and adaptability to slope variations of humanoid and biped robots, robot-assisted platforms, and neuro-rehabilitation.
Minimal feedback (once per step) as opposed to current complex schemes
• Energy efficiency
• Humanlike gait over a wide range of slopes
• Improving humanoid and biped robots, and robot-assisted platforms
• Enhancing neuro-rehabilitation
MAE-1369
以色列
