The key module for autonomous mobile robots is path planning and obstacle avoidance. Global path planning based on known maps has been effectively achieved. Local path planning in unknown dynamic environments is still very challenging due to the lack of detailed environmental information and unpredictability. This paper proposes an end-to-end local path planner n-step dueling double DQN with reward-based $ϵ$-greedy (RND3QN) based on a deep reinforcement learning framework, which acquires environmental data from LiDAR as input and uses a neural network to fit Q-values to output the corresponding discrete actions. The bias is reduced using n-step bootstrapping based on deep Q-network (DQN). The $ϵ$-greedy exploration-exploitation strategy is improved with the reward value as a measure of exploration, and an auxiliary reward function is introduced to increase the reward distribution of the sparse reward environment. Simulation experiments are conducted on the gazebo to test the algorithm’s effectiveness. The experimental data demonstrate that the average total reward value of RND3QN is higher than that of algorithms such as dueling double DQN (D3QN), and the success rates are increased by 174%, 65%, and 61% over D3QN on three stages, respectively. We experimented on the turtlebot3 waffle pi robot, and the strategies learned from the simulation can be effectively transferred to the real robot.

Despite rapid progress in UAV-based infrared vehicle detection, achieving reliable target recognition remains challenging due to dynamic viewpoint variations and platform instability. The inherent limitations of infrared imaging, particularly low contrast ratios and thermal crossover effects, significantly compromise detection accuracy. Moreover, the computational constraints of edge computing platforms pose a fundamental challenge in balancing real-time processing requirements with detection performance. Here, we present MonoSeg, a novel instance segmentation framework optimized for UAV perspective infrared vehicle detection. Our approach introduces three key innovations: (1) the Ghost Feature Bottle Cross module (GFBC), which enhances backbone feature extraction efficiency while significantly reducing computational over-head; (2) the Scale Feature Recombination module (SFR), which optimizes feature selection in the Neck stage through adaptive multi-scale fusion; and (3) Comprehensive Loss function that enforces precise instance boundary delineation. Extensive experimental evaluation on bench-mark datasets demonstrates that MonoSeg achieves state-of-the-art performance across standard metrics, including Box mAP and Mask mAP, while maintaining substantially lower computational requirements compared to existing methods.