Scientists have developed a trial and error algorithm which allows robots to adapt to damage and compensate for it.
This idea for robots that can adapt like animals was printed in Nature this week, where the process was detailed.
The algorithm is for robots that work in large search spaces. Before deploying, the robot creates a detailed map of the location which "represents the robot's prior knowledge about what behaviours it can perform and their value." If it becomes damaged, then it conducts experiments based on its knowledge to find a way to compensate.
So far the algorithm has successfully worked for robots with missing or broken legs, and damaged arms.
Continue reading…
Continue reading...
This idea for robots that can adapt like animals was printed in Nature this week, where the process was detailed.
The algorithm is for robots that work in large search spaces. Before deploying, the robot creates a detailed map of the location which "represents the robot's prior knowledge about what behaviours it can perform and their value." If it becomes damaged, then it conducts experiments based on its knowledge to find a way to compensate.
So far the algorithm has successfully worked for robots with missing or broken legs, and damaged arms.
Continue reading…
Continue reading...