Monday, February 16, 2004
Four Wheels Good, Two Legs Bad :: Robotics Trends
A good discussion on why legged robots don't make sense. Basically it's everything: less robust under uncertainty, harder from a control perspective, quickly draining power, slower, faster to get worn down, etc.
I think the article should focus more on what is fundamentally wrong with the current approach. Clearly, working legged biological machines number in the billions around the world. When a robot takes a step, many motors are working to keep the robot in balance. This makes sense. When a machine is standing still in a stable position, those same motors are still working. This is why an Asimo has a 15 minute battery life, which is unacceptable.
From a biomechanics perspective, a human act very differently. We do have tiny stabilizing muscles when we are out of balance, but most of our energy when walking or running goes into what amounts to a spring system. The energy you add in one motion is given back with the next motion.
I do think that legs will be very necessary, as the article mentions, for robots to act in a human environment (with stairs everywhere, obstacles like children's toys strewn on the ground, and other mobility requirements which wheels couldn’t handle). But, much needs to be done before this approach is feasible.
A good discussion on why legged robots don't make sense. Basically it's everything: less robust under uncertainty, harder from a control perspective, quickly draining power, slower, faster to get worn down, etc.
I think the article should focus more on what is fundamentally wrong with the current approach. Clearly, working legged biological machines number in the billions around the world. When a robot takes a step, many motors are working to keep the robot in balance. This makes sense. When a machine is standing still in a stable position, those same motors are still working. This is why an Asimo has a 15 minute battery life, which is unacceptable.
From a biomechanics perspective, a human act very differently. We do have tiny stabilizing muscles when we are out of balance, but most of our energy when walking or running goes into what amounts to a spring system. The energy you add in one motion is given back with the next motion.
I do think that legs will be very necessary, as the article mentions, for robots to act in a human environment (with stairs everywhere, obstacles like children's toys strewn on the ground, and other mobility requirements which wheels couldn’t handle). But, much needs to be done before this approach is feasible.
Comments:
Post a Comment
