Leaping 0.4m and Leaping 0.8m
In this section, we show the performance of teleoperation by overriding the arm action. In this out-of-distribution test, we show that the parkour policy can be robust to the teleoperation, even though the arm movment is not trained in the parkour task.