The first proposed strategy restricts the velocity of the end-effector not only according to the distance between human and robot, but also considering the displacement vector and the end-effector velocity vector. This is because even though the velocity restriction when the robot is moving towards the human should be firmly restricted, it is possible to relax such restriction when the robot is moving away from the human. Thus, by considering the angle between the displacement vector and the end-effector velocity vector, it is possible to favor the trade-off between human safety and efficiency.
The second proposed strategy aims to increase the distance between human and robot when such distance is too short, as increasing the distance can be naturally considered to decrease the possibility of a collision. Therefore, this strategy uses two virtual forces to make the robot move the end-effector not only away from the human but also towards a parking position.
Both strategies were tested with a human-size humanoid robot and a human standing next to each other, with a shared workspace. The results show that the proposed methods can be used for a safe interaction between human and robot.