Human motor control is a complex process, and undergoes changes due to the environmental interactions in physical human-robot interaction (pHRI). This pilot study aims to explore whether human motion under robotic constraints still complies with the same principles as in unconstrained situations, and how humans adapt to non-biological patterns of robot movements. Two typical modes in applications of pHRI (e.g., robot-assisted rehabilitation) are tested in this study. In humandominant mode, by building spring-damper force fields using a planar rehabilitation robot, we demonstrated that participants’ actual motion in reaching movements complied well with the standard minimum-jerk trajectory. However, when the virtual impedance between human force and virtual display was different from the human-robot physical impedance, the actual motion was also in a straight line but had a skewed bell-shaped velocity profile. In robot-dominant mode, by instructing participants to move along with the robot following biological or non-biological velocity
patterns, we illustrated that humans were better adapted to biological velocity patterns. In conclusion, minimum-jerk trajectory is a human preferred pattern in motor control, no matter under robotic force or motion constraints. Meanwhile, both visual feedback and haptic feedback are critical in human-robot cooperation and have effects on actual human motor control. The results of our experiments provide the background for modeling of human motion, prediction of human motion and trajectory planning in robot-assisted rehabilitation.