No abstract

No abstract

This paper presents a novel deep reinforcement learning-based system for 3D mapless navigation for Unmanned Aerial Vehicles (UAVs). Instead of using a image-based sensing approach, we propose a simple learning system that uses only a few sparse range data from a distance sensor to train a learning agent. We based our approaches on two state-of-art double critic Deep-RL models: Twin Delayed Deep Deterministic Policy Gradient (TD3) and Soft Actor-Critic (SAC). We show that our two approaches manage to outperform an approach based on the Deep Deterministic Policy Gradient (DDPG) technique and the BUG2 algorithm. Also, our new Deep-RL structure based on Recurrent Neural Networks (RNNs) outperforms the current structure used to perform mapless navigation of mobile robots. Overall, we conclude that Deep-RL approaches based on double critic with Recurrent Neural Networks (RNNs) are better suited to perform mapless navigation and obstacle avoidance of UAVs.

No abstract

No abstract

No abstract

No abstract

Autonomous underwater vehicles are a prominent tool for underwater exploration because they can access dangerous places avoiding the risks for the human beings.However, the autonomous navigation still a challenge due to the characteristics of the environment that decrease the performance of the sensor and the robot perception.In this context, this paper proposes a loop closure detector addressed to the simultaneous localization and mapping problem at semistructured environments using acoustic images acquired by forward-looking sonars. The images are segmented by an adaptative approach based on the acoustic beams analysis. A pose-invariant topological graph is build to represent the relationship between image features. The loop closure detection is achieved using a graph comparison. The approach is evaluated in a real environment at a marina. The results reveal all loop closures of the data set are detected with a high precision and present an invariant to image rotation.