First Experiments Back in the Lab

Cassie took some time off while Oregon State University halted in person research. Recently, we have been approved to return to running experiments with strict social distancing, PPE usage and careful sanitation practices. This is from our first day resuming walking tests on Cassie. The controller we are testing is a recurrent neural network policy that was trained in simulation and transferred to hardware. This is the first recurrent policy to work on Cassie that was trained without a reference motion matc