Exploration-focused training lets robotics AI immediately handle new tasks

Exploration-focused training using the MaxDiff RL algorithm enables robotics AI to handle new tasks immediately by encouraging robots to be randomly adventurous and experience a wide range of states.