OpenAI is an AI nonprofit organization founded by many Silicon Valley tycoons such as Elon Musk. Some of the warehouse robots developed are learning how to do housework and are committed to letting warehouse robots do the housework for you. It was only for artificial intelligence research at the time of its inception, and now it is reprogramming the robots produced by the FETCH robotics. Fetch Robotics is a company that provides automated warehouse facilities. OpenAI's research is to equip these robots with software that allows them to learn autonomously. OpenAI's research shows that, rather than breaking through hardware, innovation in the field of software and machine learning can better enhance the robot's capabilities. The range of warehouse robots offered by Fetch Robotics includes a robotic system that can follow workers around the factory, which transports items to a designated location. The OpenAI system is based on a mobile base station and features a three-dimensional depth sensor, a two-dimensional laser scanner, and a seven-DOF robotic arm. After intensive learning, Fetch robotics robots can do some housework. As early as April, OpenAI the Pieter Abbeel professor. Professor Pieter Abbeel is the chief expert in machine learning in the University of California, Berkeley. Professor Abbeel shows that robots can learn some of the most difficult things to do in humans, such as folding towels and taking things from refrigerators, by means of machine learning called deep reinforcement learning. DeepMind, a subsidiary of Google, which specializes in artificial intelligence, is also the way to make computers play computer games that exceed normal levels. Professor Abbeel's robot Learning project is starting from scratch. Professor Abbeel set up a neural network to receive the built-in sensor signals to control the movement of the robot, the neural network can adjust its own parameters 1.1 points, directly to complete the established action. For example, robots may crawl thousands of ways as they learn how to crawl a given goal. "Once this is achieved, the economic benefits and industrial benefits are not to be underestimated," said Marc Deisenroth. Marc Deisenroth is an expert in the field of machine reinforcement learning at Imperial College London, "Imagine how the robot can not only mop the floor but also wash the dishes, iron the clothes, clean the windows and make breakfast." ” Deisenroth says that the existing robots can reduce the cost of use by updating the software. "At the moment, software seems to be a tricky issue," he added. However, aside from this problem, upgrading the software can bring the performance of improving the robot. , he said, just as flexible arms and elastic feet are similar to monkeys ' feet, the idea is a subject that researchers are studying. Some manufacturers, along with Japan's Fanuc company, are also using intensive learning methods to train robots in factories so that robots can quickly adapt to new tasks, such as grabbing unfamiliar objects. And when the robot is trained together, the training time will be shortened accordingly. Jens Kober, from Delft University of Technology in the Netherlands, is an expert in the field of robotics, noting that "the shift from manual programming to robot-based learning is an important part of future robotics." He also pointed out that it is vital for robots to share what they have learned with each other. While robots like the fetch robotics have already been used in many factories and warehouses, the family's robot stewards remain in science fiction. It is still extremely difficult to get a machine to do seemingly simple tasks such as washing dishes, folding clothes or tidying up a room. Robots programmed in the traditional way can be paralysed in the presence of unfamiliar objects or subtle changes in light. OpenAI confirmed that they were cooperating with fetch robotics, but declined to comment. China Robot Information Network Website: http://www.jqrxx.net Information Hotline: 400-6638-111 Add: Room 1118, New era, Zijin Road, Zhengzhou, Henan, China
|