Trajectory Design Using a Behavior Model, Enabling User-Friendly Adjustment or Selection

This paper proposes a behavior model for planning trajectories suitable for even untrained laypeople to adjust the motion of their robot in real time. The method uses dynamic trajectory planning, including LQR control to track a changing target position. The adjustment of two relatable behavior trait parameters results in very different trajectory profiles. One behavior trait determines the shape of the trajectory; increasing it makes the trajectory more predictable by prioritizing movement at constant velocity, while lowering it makes the trajectory smoother and more adaptable to a changing target. The other behavior trait simply determines the amount of time the trajectory takes to reach the nominal target position. This approach should make the robot more user-friendly, since an operator can tune the motions of the robot based on changing situations or for personal preference. Moreover, mapping the robot's motion qualities to a popular understanding of human behavior enables intuitive understandings. Simulations show the trajectory generation for a single robot link, including scenarios where the target is slowly moving toward or away from its original position at a constant velocity.