Deep Reinforcement Learning for Visual Navigation and Active Tracking

In this contribution, we present our research line on Deep Reinforcement Learning approaches for robot navigation, in particular: Target-Driven Visual Navigation and Visual Active Tracking. We assess our methods capabilities in several challenging scenarios and in a number of environments previously unseen during training. Finally, we also prove that they can be effectively deployed in real-world settings on real platforms.https://youtu.be/yuFUo1gYGs