The 2023 International Robot Exhibition is being held at the Tokyo Big Sight, where a variety of robots incorporating artificial intelligence (AI) technology are no longer just following pre-programmed actions, but are learning on their own, making their judgments, and collaborating to complete tasks. These AI robots, which mimic human beings in their thinking, demonstrate the greater possibilities that AI technology will bring to the development of robotics in the future.
YASKAWA ELECTRIC, the Japanese industrial robot manufacturing giant, brought the newly released MOTOMAN NEXT series of self-regulating robots. Unlike conventional instructional reproduction robots that repeat specified actions, this series of robots can judge, plan, and execute on their own based on the surrounding conditions, and the robots can also automatically check the work situation at the right time to complete the work most appropriately.
The four self-regulating robots on display at the site demonstrated to the audience such operations as cleaning up leftovers from cutlery, removing medical instruments from trays to load baskets and feed them into sterilization equipment, and boxing up potatoes and carrots. The staff explained that in the past when robots handled different shapes of objects, they needed to be instructed one by one, and the robot’s arm needed to be programmed to move in which direction and for how much distance. With the help of AI technology, the robot can automatically generate the most suitable path for irregularly placed items, and different types of items mixed can be correctly identified, if it receives a new instruction during the operation, it can switch to a high-priority operation according to the situation, and if there is a waiting time, the robot will carry out other operations in the interval of waiting.
Shunsuke Katsuya of the public relations department of Yaskawa Electric said that it had previously been difficult to introduce automation in some fields, such as agriculture, and that increased robot autonomy could help automate these fields, thereby solving the labor shortage and the resulting social problems.
At the Waseda University booth, a robot that was hanging clothes on a hanger attracted many visitors to stop by. Students from Waseda University’s Faculty of Science and Technology, who participated in this research, told reporters that hanging clothes is trivial for people, but difficult for robots because when the robot grasps a part of the clothes, there is no way to know what shape the clothes will take in the next instant. With the help of deep predictive learning, they solved this challenge.
In the past, machine learning required constructing the most appropriate predictive model with a large amount of data in advance, but in fact, robots often encounter unexpected situations, and it is very difficult to construct a predictive model that can cope with all the situations in advance. Deep predictive learning is based on the premise of the incompleteness of the prediction model. It uses an algorithm that minimizes the prediction error between the field condition and the model so that the robot can tolerate the difference between the learning time and reality, and continue to adjust its actions in real time to achieve the goal of being able to flexibly respond to unlearned conditions.
According to reports, this robot of Waseda University can now scramble eggs, fold clothes, and do other chores, the goal of this project is to develop a symbiotic intelligent robot with people by 2050, so that the robot will become a lifelong companion of each person.
With AI technology, robots cannot only think in a way that is closer to humans, but they have also learned human teamwork. Kawasaki Heavy Industries demonstrated an integrated intelligent system featuring Nyokkey, a social robot that can move autonomously, and FORRO, a social robot for automated delivery, in a scenario where collected patient samples are sent to a lab for testing. Nyokkey, powered by a small hydrogen fuel cell, opens and closes doors, rides in lifts, etc., and places test tube racks full of sampling tubes on FORRO’s shelves, which then delivers the samples to the laboratory. In this system, Nyokkey is the mastermind, and if a sample cannot fit in one FORRO, Nyokkey will deploy another FORRO to help, and the FORROs can also communicate and collaborate.
Leave a comment