Apr 10 2007 8:40AM | Permalink | Email this | Comments (0) |
Blog This! using: Blogger.com | LiveJournal |
Even though biped robots get better and better all the time, many don't exactly have truly natural leg movements. One reason why comes down to the fact that robot legs, with the exception of a few models that employ passive locomotion, use control algorithms and joint-actuation methods that don't really approximate how a human leg works. These robots tend to move their legs according to force trajectories or pre-programmed positions, which can cause instability when the robot encounters environments and obstacles not accounted for by its control algorithms.
Shane Migliore, a researcher and engineering PhD candidate at Georgia Tech, believes that there's some room for improvement when it comes to robotic locomotion. He's been hard at work on a two-degree-of-freedom, full-scale robot leg that has dynamics much like a human leg. “We let the leg to go essentially where the leg wants to go,” he says.
Sounds simple, but it isn't. Human-like locomotion requires a mix of complex joint movements--some needing actuation, others needing a more passive approach. Migliore gives the forward swing of a human leg as a prime example. “It's almost entirely passive with very little contribution from the muscles,” he says. This mix of active and more passive movements, in turn, has created some actuation challenges for Migliore. Whereas traditional robots legs benefit from stiff joints capable of error-free actuation--the better to achieve their precisely calculated trajectories and positions--Migliore's robotic leg actually takes the opposite approach and has somewhat compliant joints. His biggest design challenge has been finding motors and actuators strong and stiff enough to deal with the forces required to move a human-scale leg while still providing enough compliance for the leg to swing naturally and respond to unforeseen obstacles. One key to the project has been using brushed DC motors from Thin Gap. They've eliminated cogging problems that make robot legs seem to index though its cycle of movement rather than swing.
Migliore eventually intends to build a complete biped robot. For now, though, he continuing work on the fundamentals of leg movement, which have implications for both a robot's overall energy usage and its ability to navigate unfamiliar environments.
For more on Migliore's leg design check out this Design News podcast.
Related entries in: Design News | Technologies |