Autonomous Robotics

Autonomous robotics is the area of ​​robotics that develops robots capable of moving and acting without human intervention. To do this, the robot must perceive its environment and act appropriately, in addition to carrying out its task.

Robotics has made great strides in structured environments, where the robot controller can have a detailed map of its environment. As the degree of structuring of the environment decreases the tasks become more complex. This happens when the robot is mobile and must have information of its position on the internal map. The mechanisms can be absolute or relative, for example using GPS and odometry, respectively.

In unstructured environments, the solution through a map is not viable, so we take paths that do not use classical artificial intelligence, with centralized control, but artificial intelligence based on multiagents (originated in the work by Rodney Brooks and its subsumption architecture), or in connectionist approaches using neural networks. The discipline that uses genetic algorithms to evolve neural networks is called evolutionary robotics.

wiki