"Put 3 fried chicken on the right of cabbage" New technology that makes Mitsubishi Electric facilitate robot introduction (ASCII) --Yahoo! News

delivery

Mitsubishi Electric has announced the development of "Teach Les Robot System Technology," which makes robots easier in fields such as food manufacturing and logistics.With high accuracy, voice recognition AI that understands the intentions of speakers and automatic optimization of multi -joint -type robot operation, even workers who do not have specialized knowledge can easily generate robot operation programs and work equivalent to humans.It is possible to achieve speed.[See more photos]

Photo: Asky

　On February 28, 2022, Mitsubishi Electric announced the development of the "Turric Les Robot System Technology" that facilitates robots. Incorporating technologies such as AI that understands the speaker's intentions with high accuracy and automatic optimization of multi -joint -type robot operation, even workers who do not have expertise can easily generate robot operation programs, and equivalent to people. It is said that the work speed can be achieved. According to the company, this is the first industrial robot maker to develop work instruction technology. Mr. Toru Oka, director of the Mitsubishi Electric Technology Research Institute, said, "The main target is food and logistics fields, and food factories that change menus frequently, such as food factories, such as food factories that have been difficult to introduce robots and sorting. It can contribute to promoting automation in the work process. " It is planned to be practical in 2023. The tingling less -robot system technology announced this time is "Easy Program generation / easy -to -adjust technology" and "Automatic optimization technology of robot operation", which was announced this time with the robot launch time without programming. Has a characteristic. This makes it easier to launch and operate robots than before, and to speed up the work to the extent that it is comparable to human workers. One of the main targets of the application is to cite the food manufacturing industry because the ratio of robot utilization in the field is still low. Mitsubishi Electric Technology Research Institute's Autonomous Control System Development Project Development Project Manager Masanoto Seki said, "In recent years, the demand for automation has increased due to labor shortages in recent years," there is a major introduction hurdle there. He points out that there is. "The background (low -utilization rate in the field of food) is that it is difficult for beginners to prepare for robots and preparing programs, time to launch and change varieties (such as changing lunch menus). There are reasons such as the need for a system that requires a system, and that human work is faster even if the system is built. " Specific systems are arm robots, controllers that move them, three vision sensors that perform static environmental scanning, mucational object scan / indefinite -specific recognition / static environmental scan, a power sensor that controls robot hands, and a personal computer that controls them. It is configured. In addition, the work instructions by the worker can enter using a 3D sensor camera and tablet. For example, on the production line of lunch boxes, a 3D sensor (or a camera at the tip of the arm robot) is scanned from various angles around the robot, creating a three -dimensional model in a virtual space. Understand the work area of ​​the arm robot, the position of the goods, the obstacle, etc. After that, use a vision sensor that performs a dynamic scan to accurately grasp the lunch box flowing on the line, follow the work instructions to operate the arm robot correctly in real time, and pack the side dishes. Work instructions to the robot can be used with a tablet and can be made with voice or simple items. In the case of audio input, you can speak, "Fill three fried chicken in the first (parcel) of the lunch box". If you want to input by entering the item, you only need to touch and select items such as "what", "Where", and "how many" in the tablet. In addition, the trajectory in which the arm robot operates by these instructions can be displayed / confirmed as a video on the tablet, so that an unexpected operation can be prevented. In the past, it was necessary to control all of these robots by programming, so it took time to make the optimal operation. By using this tyriting technology, you can automatically create a robot operation program with simple instructions, just as teaching human workers a new work. According to the company, the work time required for program generation and adjustment can be reduced to "1/10 or less". In addition, the recognition rate is enhanced by adopting audio recognition technology that is strong in the noise around the work environment, and by adopting the "intentional understanding AI model", the work content instructed by workers in natural words is high accurate. Can be estimated. It is also possible to specify an ambiguous instruction such as "a little more right" or at the relative position such as "the right of cabbage". Mitsubishi Electric's AI technology "Maisart" is characterized by a small size, and can be mounted on edge devices such as tablets. Another feature of the same technology is to automatically optimize the movement of the arm and hand, and speed up the work. For example, simply indicating the starting point and end point of the operation, the orbit that does not interfere with the surrounding facilities is automatically generated at the second level without testing. At this time, the trajectory that can maintain the maximum speed of the maximum speed and the maximum speed of the permissible torque is calculated, and the operation of the optimal speed / deceleration pattern, and the hand movement of stability while minimizing the stop time is automatically calculated. To do so, the operating time of the robot is reduced. In addition, it is a simulation of the irregular object stacking state by simulation, and it has a "grip -perception AI" that uses a trained model, and infuses the hand -grown position by high speed and high accuracy. Even in ingredients that do not have a shape like fried chicken, the angle of the arm and the opening width of the hand are determined from the image, and the balance depending on the ingredients and the balance depending on the position of the ingredients, etc., so it is stable. You can work fast. These technologies have achieved the same work speed as human work. "The work time for the robot to grasp the members and put them in the specified location was the same as 2 seconds per pick, which is equivalent to the human resources. It has been reduced "(Mr. Seki) and develops the" ROS-EDGECROSS linkage function ", and makes it easy to monitor and analyze the entire production line with the connection and multi-bender characteristic of Edgecross. It can contribute to gender and quality improvement. Mr. Seki explained, "By collecting all system logs, including ROS (Robot Operating System), with Edgecross, it can be easier to evaluate line performance." Mitsubishi Electric plans to use this technology in a sequence of food and logistics fields, as well as develop a wide range in other fields, such as electronic electronics. "In the future, we will also examine the possibility of expanding into new markets, such as supermarkets and convenience stores where multi -joint robots have not been used at all," says Seki. Sentence ● Katsuyuki Okawara Edit ● Otsuka / Tech.ascii.jp