Robotic Legs Developed with Human Walking Motion

Engineers at University of Arizona in Tucson have developed a robotic pair of legs with a biologically accurate walking motion. Theresa Klein (pictured right) and Anthony Lewis at Arizona’s Robotics and Neural Systems Laboratory published the results of their work this week in the Journal of Neural Engineering.

The researchers say the robotic legs have a simplified form of the neural and musculoskeletal architecture, and sensory feedback pathways found in human legs. The device can serve as a model to help spinal cord injury victims recover the ability to walk, as well as help researchers better understand how babies learn to walk.

The neural architecture of the the human walking system is called the central pattern generator, a neural network in the lumbar region of the spinal cord that generates rhythmic muscle signals. The human central pattern generator produces, and then controls, these signals by gathering information from different body parts in response to the environment. This neural network allows people to walk without thinking about it.

A simple form of a central pattern generator is called a half-center, which consists of two neurons that fire signals alternatively, producing a rhythm. The robotic leg system contains an artificial half-center, as well as sensors that deliver information back to the device’s half-center, including load sensors that sense force in the limb when the leg is pressed against a stepping surface. A Linux-based computer simulates the spinal neurons, receives sensory data through one line for sensory data, and commands the motors via a separate line.

Klein and Lewis demonstrated how the artificial central pattern generator helps stabilize the walking motion of the device against disruptions — such as disabling or adding extra weight to one leg — compared to a purely reflexive system, and found the Arizona device is better able to compensate for the disruptions.

The researchers also compared the trajectories of the joints in the device to human walking data, and found the robotic hips and knees resembled human walking, while the ankles were somewhat different. They attribute the difference in ankle trajectories to activation of the ankle extensors earlier in the step cycle to ensure the foot is able to clear the ground.

The following brief video demonstrates the walking motion of the robotic legs.

* * *