Robotics Basics - AI and Learning

The advances in AI and on it’s techniques has been applied to the robotics field with very good results, as the problems and tasks that have to be achieved are very complex ones that benefit from the approach that AI has on such problems. There are many techniques that have been used in the field.

Reinforcement learning

Based on feedback, the good things are repeated and the bad ones are avoided, all of them learned by trial and error. In robotics, as things are dynamic and there are errors on the measurements, the robot has to keep relearning in order to keep up to date, so there needs to be a proper balance between the exploration and the exploitation of what has been learned.

Supervised learning

There is an external “teacher” that tells what is good and bad to the robot. This is a very useful technique with the new advances on computation power and neural networks.

Learning by Imitation/Demonstration

This is a very useful technique where the robot imitates what a human does, as it can greatly simplify some behaviors programming.

Notes References

20210514183815 INDEX - Robotics

References