With the development of technology, robots begin to undertake more and more complex work. These jobs sometimes require staff to intervene at any time, so the safety of human-computer interaction is a crucial issue in this case. To ensure safety, the controller needs to detect whether there is a collision between the robot and the staff in real time, and ensure that the staff will not be injured in the collision through the control strategy. At present, most of the collision or collision force detection is achieved by adding external sensors.
There are several ways to detect collisions. The wrist force sensor is used to detect collisions: it can accurately detect the collision force at the end of the hand grip, but it cannot detect collisions at other parts of the robot. The scope of detection is limited. Usually, it is used to detect the collision force at the end of the hand grip, such as grinding force, assembly force, and so on. Perceived skin is used to detect collisions. By covering the perceptive skin over the whole body of the robot, collisions at any part can be detected. However, the disadvantages are that the wiring is relatively complex, and the anti-interference ability is poor. In the whole process, the processor has also increased the amount of computation to detect collisions through the motor current or feedback torque: this is a collision scheme widely used in various industrial robots, and no additional sensors are needed. The advantage is that the detection range can cover the entire surface of the robot.