The teleoperation feature of a leader–follower robot is used to operate the robot in a working environment that is dangerous to the operator. This paper proposes a method for estimating the intended operation of the operator of the leader–follower robot and generating behaviors for the follower robot. By generating partially autonomous robot behavior, our proposed method decreases the burden on the operator and realizes a safe controller that is robust against misoperation or disturbances. Owing to the impact of misoperation on estimation and the potential for unintended movement of the follower robot, our proposed estimation method relies on historical values of intended operation. The proposed method was verified through simulation using real operation inputs to the leader system based on practical scenarios, including misoperation.
