When performing joint tasks, the robots can maintain uninterrupted communication, such as "I passed a door and turned right 90 degrees", "Two feet forward, I met a wall, right turn 90 Degree", "4 feet forward, I met a wall"... so they can keep up with the latest developments in the event. Of course, the computer can archive a large amount of information without any difficulty, and immediately call it out when needed to judge the current situation, but for humans, such a series of intensive data will make people crazy.
At the recent annual meeting of the American Association for Artificial Intelligence Development, researchers at the Massachusetts Institute of Technology's Computer Science and Artificial Intelligence Laboratory (CSAIL) proposed a new approach to simulating robotic collaboration that can reduce the amount of communication required. %. The new model makes it easier for designers and robots to work together, such as creating a human-machine emergency rescue team.
Reduce multi-agent system communication costs
In computer terminology, the Multiagent System is a system that enables arbitrary collaboration between autonomous agents (electronic or non-electronic). Each subject must be able to run a model that reflects the current state of the world it recognizes, the robot worldview, and the worldview model of any other robot. In the model accuracy probability, the robot itself is also a factor to be considered, and they must be based on these probabilities to decide whether to change their behavior.
The method of simulating a multi-agent system is called Dec- POMDP (Distributed Partially Observable Markov Decision Process). The Dec-POMDP system includes a variety of uncertainties. It is necessary to consider whether the world view of a certain robot in the team is correct, whether it is correct in determining the partner's worldview, and whether any action taken by it can be successful. For example, the robot was originally planned to move 20 feet forward, but found that the wind blowing in the horizontal direction blows itself off the route. If a robot decides to release a new message, it will force the companion to update its model and stir up a new round of uncertainty in the whole. If this information is not necessary, it will only cause serious stagnation and will not achieve any purpose.
In general, the Dec-POMDP system makes some a priori assumptions about the environmental conditions, allowing the robot to act on these assumptions. However, the emergency response system cannot make assumptions first, because the rescue team usually enters an unfamiliar environment, and the best prior knowledge does not work. Moreover, emergency rescue requires a comprehensive grasp of the environment, and it is extremely time-consuming to calculate a multi-robot solution. The researchers were designed to let the system ignore the uncertainty of the action, assuming that no matter what a robot is going to do, it will do it.
Coordinate individual actions to optimize the overall
When the robot receives a new message, such as a passage in a building is closed, it has three choices: one is to ignore this information; the second is to adopt it but not to publish it; the third is to adopt and publish the information - Each choice has advantages and corresponding communication costs. If a robot incorporates new information into its own worldview model but does not issue it, it will also incur costs, that is, it will be more difficult for a companion to judge its worldview. For each new piece of information acquired by each robot, the new system performs a cost-benefit analysis based on their world model, their expectations of partner behavior, and the likelihood of more common goals being achieved.
Open Frame Power Supply,Led Driver Power Supply,175W Open Frame Power Supply,150W Open Frame Power Supply
Guang Er Zhong(Zhaoqing)Electronics Co., Ltd , https://www.cnadaptor.com