Researchers at the Hybrid Robotics Group at UC Berkeley and CMU are hard at work making sure their robots don’t fall over when tiptoeing through rough terrain. Using machine learning and ATRIAS robots, the teams are able to “teach” robots to traverse stepping stones they’ve never seen before.
Their robots, described here, are unique in that they are bipedal and use a mixture of balance and jumping to ensure they don’t tip off the blocks.
“What’s different about our methods is that they allow for dynamic walking as opposed to the slower quasi-static motions that robots tend to use,” write the researchers. “By reasoning about the nonlinearities in the dynamics of the system and by taking advantage of recent advances in optimal and nonlinear control technology, we can specify control objectives and desired robot behaviors in a simple and compact form while providing formal stability and safety guarantees. This means our robots can walk over discrete terrain without slipping or falling over, backed by some neat math and some cool experimental videos.”
The robots are currently “blind” and can’t use visual input to plan their next move. However, with a robot called CASSIE, they will be able to see and feel the stones as they hop along, ensuring that they don’t tip over in the heat of fun… or battle.