<ul><li>This work proposes a novel method, based on the Reinforcement Learning (RL) paradigm, to train redundant robots to be able to execute multiple tasks concurrently.</li><li>The approach allows combining and executing learned tasks in possibly time-varying prioritized stacks.</li><li>The method defines task independence between learned value functions and uses a cost functional to encourage the accomplishment of control objectives while minimizing interference with higher priority tasks.</li><li>The authors demonstrate the effectiveness of the approach on several scenarios and robotic systems.</li></ul>

Value Iteration for Learning Concurrently Executable Robotic Control Tasks

Discover more