Abstract
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for controlling a robot based on settling time of the robot. One of the methods includes executing, by a robot having one or more joints, a motion trajectory to move from a first pose to a second pose; repeatedly computing a settled state confidence score after the robot has begun execution of the motion trajectory; and whenever the current settled state confidence score satisfies a threshold value, executing, by the robot, a command that causes the robot to move.