Abstract:In order to solve the problem of frequent deceleration of unmanned vehicles when approaching obstacles, this article uses a Deep Q-Network (DQN) and its extension, the Double Deep Q-Network (DDQN), to develop a local navigation system that adapts to obstacles while maintaining optimal speed planning. By integrating improved reward functions and obstacle angle determination methods, the system demonstrates significant enhancements in maneuvering capabilities without frequent decelerations. Experiments conducted in simulated environments with varying obstacle densities confirm the effectiveness of the proposed method in achieving more stable and efficient path planning.

What problem does this paper attempt to address?

This paper aims to solve the problem of frequent deceleration of unmanned vehicles when approaching obstacles. Specifically, the author uses the Deep Q - Network (DQN) in deep reinforcement learning and its extended version, the Double Deep Q - Network (DDQN), to develop a local navigation system that can adapt to obstacles and maintain optimal speed planning. By introducing an improved reward function and an obstacle angle determination method, this system significantly improves the vehicle's handling ability, reduces unnecessary deceleration, and thus achieves more stable and efficient path planning. ### Main Contributions 1. **Coupling Speed and Obstacle Angle**: The paper proposes to couple the vehicle speed with the angle between the vehicle and the obstacle and integrate this coupling relationship into the reward function to improve the flexibility and efficiency of the system. 2. **Using the DDQN Algorithm**: The DDQN algorithm is used to replace the traditional local path - planning module to reduce the over - estimation problem of Q - values and improve the stability and reliability of learning. 3. **Dynamic Reward Function**: A dynamic reward function is designed to dynamically adjust the reward value according to the current speed of the vehicle and the angle relationship with the obstacle, so that the vehicle can maintain an appropriate speed when approaching the obstacle without having to decelerate frequently. ### Experimental Verification The paper conducted experiments in the Gazebo simulation environment to test the vehicle speed - planning performance in different environments. The experimental results show that the improved reward function can achieve a more stable driving speed in environments with various obstacle densities and reach the expected performance goals. ### Conclusion By improving the reward function and using the DDQN algorithm, this research has successfully improved the speed - planning ability of unmanned vehicles in complex environments, reduced unnecessary deceleration, and improved driving efficiency and safety. These methods have been verified in simulation experiments, demonstrating their potential in practical applications.

Adaptive speed planning for Unmanned Vehicle Based on Deep Reinforcement Learning

Path Planning Technology of Unmanned Vehicle Based on Improved Deep Reinforcement Learning

Spacecraft Attitude Maneuver Planning Based on Deep Reinforcement Learning under Complex Constraints

A RDA-Based Deep Reinforcement Learning Approach for Autonomous Motion Planning of UAV in Dynamic Unknown Environments

Deep Reinforcement Learning Approach with Multiple Experience Pools for UAV's Autonomous Motion Planning in Complex Unknown Environments

UAV Coverage Path Planning Based on Deep Reinforcement Learning

Improved Robot Path Planning Method Based on Deep Reinforcement Learning

Unmanned Aerial Vehicle Path Planning Algorithm Based on Deep Reinforcement Learning in Large-Scale and Dynamic Environments

Path Planning for Autonomous Vehicles in Unknown Dynamic Environment Based on Deep Reinforcement Learning

Obstacle avoidance planning of autonomous vehicles using deep reinforcement learning

Autonomous Navigation of Unmanned Vehicle Through Deep Reinforcement Learning

Prioritized experience replay-based DDQN for Unmanned Vehicle Path Planning

Path Planning of Autonomous Mobile Robot in Comprehensive Unknown Environment Using Deep Reinforcement Learning

Autonomous Navigation of UAV in Multi-Obstacle Environments Based on a Deep Reinforcement Learning Approach

Risk-aware Deep Reinforcement Learning for Decision-making and Planning of Autonomous Vehicles

A UAV Path Planning Method Based on Deep Reinforcement Learning

Unmanned Aerial Vehicle Path Planning Based on Improved DDQN Algorithm

An Improved Deep Reinforcement Learning Algorithm for Path Planning in Unmanned Driving

Autonomous obstacle avoidance of UAV based on deep reinforcement learning

Based on Doubly Decoupled Reinforced Network

Design and Experimental Validation of Deep Reinforcement Learning-Based Fast Trajectory Planning and Control for Mobile Robot in Unknown Environment