Robot collision detection system and method based on neural network

Abstract

The invention relates to a robot collision detection system and method based on a neural network. The method comprises the steps that the actual position of each joint motor is detected and recorded in each instruction period of a robot; the instruction position, the instruction speed and the actual position of each joint motor are taken as input characteristic data of the neural network to predict the position of the joint motor at the current moment; the deviation between the measured actual position at the current moment and the predicted joint motor position is compared, and if the deviation exceeds a set range, it is judged that collision occurs; and otherwise, it is judged that no collision occurs. According to the system and method, no additional hardware equipment is needed, and collision detection can be realized only by a position sensor of the robot. Meanwhile, kinetic parameters of the robot do not need to be known. Collision can still be accurately detected under the working condition that the load of the robot changes, and misjudgment cannot be caused in the acceleration and deceleration process of the robot. The detection method has the advantages of being high in collision detection precision, concise in algorithm, simple in calculation and good in universality.

Description

technical field [0001] The invention relates to the technical field of positioning methods, in particular to a neural network-based robot collision detection system and method. Background technique [0002] With the development of science and technology, the application of robots has penetrated into all aspects of human society. Industrial robots can replace workers to perform repetitive physical labor efficiently and accurately. Workers have the advantage of being more flexible than robots. Therefore, workers and robots working together can greatly improve work efficiency. Collaborative robots should face unpredictable possible collisions in irregular and dynamically changing environmental conditions to ensure the safety of workers and robots themselves. [0003] An important topic that must be broken through in the development of the human-machine collaborative robot industry. In the production process, human-robot collaboration can allow robots to better cooperate wit...

AI Technical Summary

Problems solved by technology

[0010] (1) The scheme of judging the collision based on the joint motor current or torque step peak is simple and easy to implement, but it is very easy to cause misjudgment

During the acceleration and deceleration process of the joint motor, the current or torque of the joint motor will also have a step, which is very easy to cause the system to misjudge that the robot has collided.

[0011] (2) The main defects of the scheme based on robot inverse dynamics are two aspects: a. The scheme is very sensitive to the dynamic parameters of the robot (joint position, joint velocity, joint acceleration, inertia, connecting rod mass, friction, etc.), and these Many parameters are time-varying and not easily obtained

c. In the measurement of actual torque, additional hardware devices such as force sensors need to be added (although some solutions can observe torque through algorithms, there is a large error between the observed torque and the actual torque, and the accuracy of observation is far inferior to the actual measurement of the sensor )

[0012] (3) The scheme based on the skin sensor (pressure sensor) can detect the collision accurately, but this scheme increases the complexity of the robot system, reduces the flexibility of the robot, and greatly increases the cost of the system. poor promotion

[0013] (4) The main defects of the vision sensor-based solution include: a. There is a blind spot in the vision sensor

b. At present, the precision of industrial robots is very high (repeated positioning accuracy is usually 10 -2 mm order of magnitude), the accuracy of the scheme of detecting whether physical contact has occurred according to the visual sensor may not meet the requirements of collision detection at all

c. The data volume of the visual sensor is huge, the processing time is long, and it is difficult to meet the real-time requirements of collision detection

[0014] (5) The scheme based on the energy or generalized momentum of the robot system and the scheme based on the inverse dynamics of the robot have the same defects as the scheme based on the inverse dynamics of the robot

Images