A system has appeared that helps robots navigate even in tight spaces. In doing so, it relies not on maps or GPS, but on the analysis of the environment by the device.
The researchers explained that there are already several systems and techniques for navigating robots, but their mobility is limited, devices are lost in environments unknown to them. Most existing navigation methods have two main components: one is for building a map that the robot can use as a guide, and the other is a path generator.
Researchers at Nanjing University of Aeronautics and Astronautics and China’s National University of Defense Technology have developed a new system that can provide more efficient indoor robotic navigation. Rather than relying on maps, this system uses a training approach known as simulation learning, which allows robots to navigate their surroundings.
“Our method provides for multiple observation of the robot, adjusting the input data so that it does not rely on a map or GPS in the future. Instead, he independently adjusts to the space, analyzing it, ”the scientists note.
The navigation system, developed by the researchers, has three key components. The first is a variational generative module, trained on human demonstrations, which is designed to predict changes in the environment. The second component predicts static collisions, enhancing the safety of the robot’s navigation. Finally, the Goal Verification module looks at the end action or goal that the robot is trying to achieve.
In the future, the new system can be used to improve the navigation of robots designed for home, office or other environments.