Add to Favourites
To login click here
Fido is an AI that allows robots to learn organically over their lifetime, eliminating the need for an expert programmer. It uses a modified version of the Q-learning reinforcement learning algorithm, coupled with a neural network and a wire-fitted interpolator. Fido was tested on three robots of varying processing capability and kinematics, and was able to successfully learn various tasks in simulation and in hardware with low latency. It is 4x faster than the industry standard, making it practical for use.