The quadruped robot looks a little bit like a headless, medium-size dog. Photograph by Michael Price, courtesy of SDSU.
Kaveh Akbari Hamed, PhD, an assistant professor in the San Diego State University (SDSU) Department of Mechanical Engineering, has been working on autonomous robots for the past ten years with the aim of helping people with prosthetic legs walk more naturally. More recently, Hamed has been working with a quadruped robot named SQ1, the result of a recently awarded three-year, $612,000 grant from the National Science Foundation.
Hamed is using SQ1 and other robots like it to write and test local-control algorithms that make the robots walk naturally and efficiently. Whereas most approaches to robotic leg movement rely upon a centralized computer system that monitors and controls each limb, he is working on a decentralized system. Ideally, each leg would sense its own pressure and angle and respond accordingly to maintain a balanced gait.
“It’s based on the biology of animals,” Hamed said. “Studies have shown that in animals, locomotion is decentralized. Each leg monitors its own feedback and has its own controllers.”
While the brain directs higher-level functions, like turning or stepping over a crack, the simple act of walking while keeping your balance is handled primarily by signals sent between the leg muscles and spinal cord-in other words, local control. The benefit of making prosthetic legs with local control is that each leg operates independently, reducing the chances of a system-wide crash, as compared with a centralized control system. It’s also more energy efficient, meaning batteries can last longer.
“But mathematically, the problem is very complicated,” he said.
As Hamed develops these local-control algorithms at SDSU, he hands them off to his collaborator at the University of Texas at Dallas, Robert Gregg, PhD, an assistant professor of mechanical engineering and bioengineering, who tests them with a powered prosthetic leg for humans. Ultimately, their goal is to imbue the prosthetic leg with a balancing algorithm that makes walking with it effortless, automatic, and energy efficient.
“Right now, the simulation results are promising,” Hamed said. “We want to translate that into reality.”
Editor’s note: This story was adapted from materials provided by Michael Price and SDSU.
