Perhaps most people haven’t considered the problems and hidden dangers that there are many robots in this world – you will waste a lot of time avoiding collisions between them, and what’s worse, in the process, they also The task cannot be completed normally.
At Georgia Tech, Li Wang, Professors Aaron D. Ames, and Magnus Egerstedt have been working on ways to move large numbers of robots around without getting in the way or bumping into other robots. This is crucial for users like me who have 37 Roomba Robot vacuums in their home. Of course, if in the future, the streets are full of driverless cars, this method is also very important.
The most fundamental problem to be solved in this field is Robot paranoia. When robots walk around, for the sake of safety, they will set up a “panic zone” for themselves based on sensors. Once an object from the outside world enters this area, they will be paranoid and “fear” and stop moving. Just imagine, if there are only two robots in an area, they will keep a certain safe distance from each other and not interfere with each other; but as the number of robots increases, the probability of their “panic zones” intersecting will increase. After the threshold their “panic zones” coincide. If this were to happen globally, you would be witnessing an anomaly in which robots all over the world are paralyzed. In this regard, the Georgia Institute of Technology research and development team said: “As the number of robots increases, the difficulty of the tasks to be performed also increases, so it will become more difficult to use only one controller to complete different tasks at the same time, which may involve Tasks include object shaping, collision avoidance, and connectivity maintenance.”
Robots in the traditional sense use several different controllers in the process of completing tasks. Among them, the main controller focuses on getting the Robot to complete tasks, such as obeying commands like “go there.” The second controller, the safety controller, ensures that the primary controller works properly so the robot doesn’t bump into other objects. Most of the time, the safety controller is not running, but once it senses that the robot is at risk of being hit, it will take over and control the robot’s actions instead of the main controller. At this point, the problem comes – after the safety controller is activated, the robot will always be in a “safe mode” that avoids collisions, which greatly hinders the completion of normal tasks.
To solve this problem, Georgia Tech’s research and development team designed a safety controller for mobile robots that does not interfere as much with the main controller, which means that “only in the event of a collision, or a connected In situations where damage is already imminent, the safety controller can control the robot to take action to avoid a collision.” To test their algorithm on several robots, the R&D team conducted experiments with Khepera III, a small mobile robot developed by Swiss company K-Team. Ideally, the safety controller would always function, but when you want the controller to “accomplish complex tasks” while also “avoiding collisions and maintaining connectivity”, it becomes more difficult.
Scientists believe that technologies like this will become increasingly important as driverless cars fill our streets in the near future and there will be more and more drones in the sky. The safety measures of robots are of course important, but if they cannot reliably complete their tasks while ensuring their own safety, no matter how many robots humans eventually have, their role will be very limited.
The Links: SGMRS-12A2B-YR12 AC2251
