Traditional industrial robots have a large volume and low safety factor, as there are no people allowed within the operating radius. With the increasing demand for dynamic unstructured production such as precision manufacturing and flexible manufacturing, the coexistence of robots with humans and robots with the environment has put forward higher requirements for robot design. Robots with this ability are called collaborative robots.
Collaborative robots have many advantages, including lightweight, environmental friendliness, intelligent perception, human-machine collaboration, and ease of programming. Behind these advantages, there is a very important function, which is collision detection - the main function is to reduce the impact of collision force on the robot body, avoid damage to the robot body or peripheral equipment, and more importantly, prevent the robot from causing damage to humans.
With the development of science and technology, there are many ways to achieve collision detection for collaborative robots, including kinematics, mechanics, optics, etc. Of course, the core of these implementation methods are components with various detection functions.
Collision detection of collaborative robots
The emergence of robots is not intended to completely replace humans. Many tasks require cooperation between humans and robots to complete, which is the background of the birth of collaborative robots. The original intention of designing collaborative robots is to interact and collaborate with humans in work, in order to improve work efficiency and safety.
In a work scenario, collaborative robots collaborate directly with humans, so safety issues cannot be overemphasized. In order to ensure the safety of human-machine cooperation, the industry has formulated many relevant regulations and standards, with the aim of considering the safety issues of human-machine cooperation from the design of collaborative robots.
Meanwhile, collaborative robots themselves must also ensure safety and reliability. Due to the high degree of spatial freedom of collaborative robots, which mainly replace human work in complex and dangerous environments, it is also necessary to quickly and reliably detect potential collisions in grinding, assembly, drilling, handling and other work.
In order to prevent collisions between collaborative robots and humans and the environment, designers roughly divide collision detection into four stages:
01 Pre-collision detection
When deploying collaborative robots in a work environment, designers hope that these robots can be familiar with the environment like humans and plan their own movement paths. To achieve this, designers install processors and detection algorithms with certain computing power on collaborative robots, and build one or several cameras, sensors, and radars as detection methods. As mentioned above, there are industry standards that can be followed for pre collision detection, such as the ISO/TS15066 collaborative robot design standard, which requires collaborative robots to stop running when people approach and immediately recover when people leave.
02 Collision detection
This is a either yes or no form, representing whether the collaborative robot has collided. In order to avoid triggering errors, designers will set a threshold for collaborative robots. The setting of this threshold is very meticulous, ensuring that it cannot be triggered frequently while also being extremely sensitive to avoid collisions. Due to the fact that the control of robots mainly relies on motors, designers combine this threshold with motor adaptive algorithms to achieve collision stop.
03 Collision isolation
After the system confirms that a collision has occurred, it is necessary to confirm the specific collision point or collision joint. The purpose of implementing isolation at this time is to stop the collision site. The collision isolation of traditional robots is achieved through external guardrails, while collaborative robots need to be implemented through algorithms and reverse acceleration due to their open space.
04 Collision recognition
At this point, the collaborative robot has confirmed that a collision has occurred, and the relevant variables have exceeded the threshold. At this point, the processor on the robot needs to determine whether the collision is an accidental collision based on sensing information. If the judgment result is yes, the collaborative robot needs to self correct; If it is determined as a non accidental collision, the collaborative robot will stop and wait for human processing.
It can be said that collision detection is a very important proposition for collaborative robots to achieve self-awareness, providing the possibility for the large-scale application of collaborative robots and entering a wider range of scenarios. At different collision stages, collaborative robots have different requirements for sensors. For example, in the pre-collision detection stage, the main purpose of the system is to prevent collisions from occurring, so the responsibility of the sensor is to perceive the environment. There are many implementation routes, such as vision based environmental perception, millimeter wave radar based environmental perception, and lidar based environmental perception. Therefore, corresponding sensors and algorithms need to be coordinated.
After a collision occurs, it is important for collaborative robots to be aware of the collision point and degree as soon as possible, in order to take further measures to prevent the situation from deteriorating further. The collision detection sensor plays a role at this time. The common collision sensors include mechanical collision sensors, magnetic collision sensors, piezoelectric collision sensors, strain type collision sensors, piezoresistive plate collision sensors, and mercury switch type collision sensors.
We all know that during the operation of collaborative robots, the robotic arm is subjected to torque from many directions to make the robotic arm move and work. As shown in the figure below, the protection system equipped with collision sensors will apply a combined torque, torque, and axial load reaction force upon detecting a collision, and the collaborative robot will immediately stop working.
Post time: Dec-27-2023
