Abstract:Nowadays, autonomous vehicles are gaining traction due to their numerous potential applications in resolving a variety of other real-world challenges. However, developing autonomous vehicles need huge amount of training and testing before deploying it to real world. While the field of reinforcement learning (RL) has evolved into a powerful learning framework to the development of deep representation learning, and it is now capable of learning complicated policies in high-dimensional environments like in autonomous vehicles. In this regard, we make an effort, using Deep Q-Learning, to discover a method by which an autonomous car may maintain its lane at top speed while avoiding other vehicles. After that, we used CARLA simulation environment to test and verify our newly acquired policy based on the problem formulation.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to use deep reinforcement learning techniques in self - driving vehicles, enabling the vehicles to keep driving in the lane at the highest speed in the CARLA simulation environment while avoiding collisions with other vehicles. Specifically, the author explores a strategy through the Deep Q - Learning method, so that self - driving cars can drive efficiently and safely in complex traffic environments. The main research objectives in the paper include: 1. **Improve the safety and efficiency of self - driving**: By designing appropriate reward mechanisms and reinforcement learning algorithms, self - driving vehicles can effectively avoid collisions with surrounding vehicles while maintaining high vehicle speeds, thereby improving driving safety and efficiency. 2. **Solve the problem of policy learning in high - dimensional environments**: Utilize deep reinforcement learning techniques, especially Deep Q - Learning, to solve the problem of learning complex policies in high - dimensional environments (such as self - driving scenarios). This involves how to effectively extract useful information from a large amount of sensor data and make the best decisions accordingly. 3. **Verify the effectiveness of the proposed method**: Through extensive testing in the CARLA simulation environment, verify the effectiveness and robustness of the proposed self - driving strategy based on deep reinforcement learning. This includes evaluating the vehicle's performance under different urban environments, weather conditions, and traffic densities. To achieve these goals, the author describes in detail the design of the observation space, action space, and reward structure, as well as the specific implementation methods of the multi - layer perceptron model (MLP) and DQN algorithm. Through these designs and methods, the paper shows how to use deep reinforcement learning techniques to train self - driving vehicles to perform well in complex traffic environments.

Autonomous Driving with Deep Reinforcement Learning in CARLA Simulation

Safe Navigation: Training Autonomous Vehicles using Deep Reinforcement Learning in CARLA

Deep reinforcement learning based control for Autonomous Vehicles in CARLA

Automated Driving Maneuvers under Interactive Environment based on Deep Reinforcement Learning

End-to-End Deep Reinforcement Learning for Lane Keeping Assist

Reinforcement Learning-Based Autonomous Driving at Intersections in CARLA Simulator

Autonomous Highway Driving using Deep Reinforcement Learning

Think2Drive: Efficient Reinforcement Learning by Thinking in Latent World Model for Quasi-Realistic Autonomous Driving (in CARLA-v2)

Deep Reinforcement Learning framework for Autonomous Driving

Self-Driving Car Racing: Application of Deep Reinforcement Learning

Think2Drive: Efficient Reinforcement Learning by Thinking with Latent World Model for Autonomous Driving (in CARLA-v2)

Deep Reinforcement Learning for Autonomous Driving: A Survey

Comprehensive Training and Evaluation on Deep Reinforcement Learning for Automated Driving in Various Simulated Driving Maneuvers

Safe Autonomous Driving with Latent Dynamics and State-Wise Constraints

From Naturalistic Traffic Data to Learning-Based Driving Policy: A Sim-to-Real Study

A Methodology Based on Deep Reinforcement Learning to Autonomous Driving with Double Q-Learning

An End-to-End Curriculum Learning Approach for Autonomous Driving Scenarios

Overcoming Challenges of Applying Reinforcement Learning for Intelligent Vehicle Control

A reinforcement learning based autonomous vehicle control in diverse daytime and weather scenarios

Human-like Autonomous Vehicle Speed Control by Deep Reinforcement Learning with Double Q-Learning

Applications of Distributional Soft Actor-Critic in Real-world Autonomous Driving