Reflex Control of Robotic Gait Using Human Walking Data

The study aims to simplify the control of robotic walking by using data from human walking patterns. Instead of complex neural networks, the robot's movements are controlled by reflexes triggered by ground contact, similar to human walking. Researchers measured the relationship between heel contact and muscle activity in humans while walking. This data was used to create 'transfer functions' that convert sensory data into motor actions for the robot. These transfer functions were then applied to RunBot II, a bipedal robot, resulting in a stable and controlled gait. This suggests that these functions could be used in assistive devices for gait retraining, especially for individuals with spinal cord injuries.

• 1
  There is a causal relationship between ground contact information from the heel and EMG, which allows the creation of a minimal, linear, analogue control system for controlling walking.
• 2
  The derived transfer functions, when applied to RunBot II, produced a stable and controlled gait cycle.
• 3
  Characteristics within the transfer functions related to flexion or extension of the hip and knee joints in humans were identified and separated to produce transfer functions for controlling RunBot’s leg motors and generate stepping.