Without AI, a collaborative robot is a lightweight robotic arm, and its biggest advantage compared to traditional robots will only be ‘light’. Therefore, when we designed the collaborative robot body, we also started the research and development of AI and related technologies, products, and introduced the strategic partner Dr. Sean Lee and his team to the process. According to Sun Kai, ELITE CTO, the lightweight element is the carrier of collaborative robots, but AI is the soul. The machinery, structure, and traditional control technologies are the foundation, but the core that allows collaborative robots to shine are the functions such as vision and AI.
Aiming at the problem of how AI empowers collaborative robots, the GG robotics network interviewed Dr. Sun Kai, CTO of ELITE.
■ Reinterpreting “Collaborative Robots”
Collaborative robots, as the name implies, emphasizes the concept of “collaboration”, focusing on the cooperation between humans and robots. As a result, ease of use, safety, and intelligence are especially important.
“Can all good-looking and lightweight robots equipped with modular hollow joints be called collaborative robots?” Sun Kai suggested otherwise: “There are some collaborative robots on the market that still do more traditional handling, loading and unloading, or relatively independent assembly work during the actual production process, that are not ‘1 to 1’ collaboration with people. “
“Without intelligence, you can’t really reach the stage of true collaboration.” Sun Kai firmly believed in that. In order to achieve the ambitious goal that robots have the same “intelligence” as humans, ELITE’s collaborative robots were designed with the combination of computer vision and AI technology at the very beginning.
From the perspective of the body structure, the central hole of the joint is large, which is convenient for the built-in high-speed communication lines such as USB and Ethernet. It also provides a video signal interface and a fixed point on the end flange. Each joint is equipped with a MEMS sensor that is able to obtain richer perception data.
From the perspective of software and hardware, a powerful computing platform with a GPU is provided in the control system to provide sufficient computing power for AI. In addition to providing the performance and functions of ELITE universal robot controller, it also provides the integration with the camera to achieve comprehensive support for 2D, 2.5D to 3D vision software and algorithms.
At the same time, through the introduction of deep learning frameworks and deep learning training on large-scale data, the robot itself can imitate human perception of objects and achieve intelligent capture, thereby having the ability to understand the surrounding environment and identify objects.
In addition, ELITE also provides visual and AI functional interfaces that can be operated remotely, not only retaining the convenience of operation, but also providing solutions for complex situations from the software level.
Last year, ELITE showed two demos about the future expression of “AI + collaborative robots”. In the random mixed object grabbing system, the latest deep neural network technology is applied, and basing on the self-developed object recognition and grasping pose prediction algorithms, the robot is guided to achieve accurate grabbing of random objects without having to obtain any object model data in advance.
Through the fusion of 2D and 3D image data, the system reduces the accuracy requirements of the deep vision sensor, improves the success rate of capture, enhances the versatility of the capture system, and supports the capture of mixed objects. The system can be widely used in 3C, logistics, retail, catering, and other industries to provide a solid foundation for grabbing applications for service industry robots and “smart workers” in industrial environment.
In another demo — the robot stacking clothes system, it uses the deep reinforcement learning algorithm and deep vision sensor to accurately locate the clothing stacking point, and automatically optimizes the best motion trajectory to achieve the stacking effect. The system also uses rapid simulation modeling and transfer learning methods to speed up the learning speed and reduce the cost of data acquisition, and finally map the simulation results to the actual robot operations.
Sun Kai revealed that ELITE is currently working hard on new applications of AI technology in robotic carriers. For example, teaching imitation, that is, no teaching pendant and drag teaching pendant will be needed. After a person performs an action, the robot can recognize, understand and imitate the human to perform this set of actions. Another example would be obstacle avoidance, which means that the robot will slow down or stop moving when it finds someone approaching.
In addition to the collaborative robot and AI-related research and development, whether it is an industrial or consumer-facing robot, the collaborative robot body alone cannot meet the application requirements; it also requires complex tool-side design. Sun Kai told us that ELITE is currently developing a two-finger 300N gripper for industrial environments and an under-actuated smart hand with a three-finger 30N grip for service industries to put on the self-developed collaborative robots.
In application scenarios, relying on ELITE’s advantages in controllers and vision, ELITE is able to provide customers with one-stop system integration solutions, including handling, loading and unloading, palletizing, assembly, and vision-based mixed item sorting, facing new retail such as food sales, etc.
■ Run with The Dream, “Smart Workers” on the Road
It is widely believed in the industry that with the penetration of collaborative robots into the industrial market, they can hopefully enter extend from the existing B-end market into the C-end market to provide services for families and individuals. At this time, collaborative robots are no longer just a device, but a creative tool for family life. Elliott thinks so too, and it is moving towards this dream.
“The large-scale application of collaborative robots requires more advanced control methods. This control method is different from traditional kinematics, dynamics, PID, etc. The improvement of the control method will eventually be extended to service robots.” Sun Kai said.
As of now, ELITE’s 6-axis 6 kg and 7-axis 5 kg collaborative robots have been developed and preparations for mass production have begun. The robots will officially launch in June this year, and ELITE will likely become one of the first manufacturers in China that mass-produce seven-axis collaboration. In addition, ELITE will officially launch 6-axis 3 kg and 7-axis 3 kg collaborative robots in September this year to further improve its collaborative robot product line.
The current robot is far away from what a robot looks like in the minds of the public. Therefore, the moment when the robot meets the psychological expectations of ordinary people, is when it makes its grand appearance on the stage of the era. Sun Kai believed: “The real robot should be capable of” hand-brain coordination “, with both execution and thinking capabilities. With the advancement of AI and related control technologies, the structural design of robots will also keep up with the times. The current form of collaborative robots is not ideal, and the final product form will be a human-like intelligent dual-arm robot. “
Based on this, ELITE will also launch a new series of collaborative robot products with human-like arm structure, weighing less than 5 kg, and selling at half the price of existing collaborative products. It is expected to be used in new retail, intelligent sorting, logistics, medical, education and other fields.
“The biggest opportunity for Chinese robots comes from intelligent robots. This is not only an opportunity to overtake a curve, but also a manifestation of China’s advantages and bonuses. Collaborative robots will become more and more human-like and will eventually become ‘smart workers’. They will step onto the stage of history, become a label of the era and an indispensable thing in the hearts of the public. “In Sun Kai’s view, this vision will be realized soon enough, and ELITE will continue on this path unwaveringly.
In the end, Sun Kai revealed that this year ELITE will continue to increase funding for mass production and marketing of collaborative robots and further research and development of machine intelligence and machine vision to improve the layout of its entire product matrix.