Autonomous Driving using Residual Sensor Fusion and Deep Reinforcement Learning

Amin Jalal Aghdasian,Amirhossein Heydarian Ardakani,Kianoush Aqabakee,Farzaneh Abdollahi

2023-12-27

Abstract:This paper proposes a novel approach by integrating sensor fusion with deep reinforcement learning, specifically the Soft Actor-Critic (SAC) algorithm, to develop an optimal control policy for self-driving cars. Our system employs a two-branch fusion method for vehicle image and tracking sensor data, leveraging the strengths of residual structures and identity mapping to enhance agent training. Through comprehensive comparisons, we demonstrate the efficacy of information fusion and establish the superiority of our selected algorithm over alternative approaches. Our work advances the field of autonomous driving and demonstrates the potential of reinforcement learning in enabling intelligent vehicle decision-making.

Systems and Control

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to develop an optimal control strategy in self - driving cars by integrating sensor fusion and deep reinforcement learning techniques. Specifically, the paper proposes a new method, which combines the advantages of the residual structure and identity mapping, and processes vehicle images and tracking sensor data through a two - branch fusion method to enhance agent training. This method aims to improve the decision - making ability and environmental perception ability of the self - driving system, thereby enhancing road safety and driving efficiency. The paper mentions that traditional control strategies such as PID controllers, MPC (Model Predictive Control), LQR (Linear Quadratic Regulator), etc., are effective in specific scenarios, but have limited performance in complex and dynamic environments. However, methods based on deep reinforcement learning, especially the Soft Actor - Critic (SAC) algorithm used in this paper, can continuously learn and adapt in such environments to achieve better performance. In addition, the paper also emphasizes the importance of sensor fusion. By fusing data from different sensors, a more comprehensive and accurate environmental model can be generated, which is crucial for reinforcement learning algorithms because the learning effect of these algorithms largely depends on the quality of the perceived information. Therefore, this research not only promotes the technological development in the field of self - driving, but also shows the potential of reinforcement learning in enabling intelligent vehicle decision - making.

Autonomous Driving using Residual Sensor Fusion and Deep Reinforcement Learning

Exploiting Multi-Modal Fusion for Urban Autonomous Driving Using Latent Deep Reinforcement Learning

Multi-modal policy fusion for end-to-end autonomous driving

Joint Optimization of Sensing, Decision-making and Motion-controlling for Autonomous Vehicles: A Deep Reinforcement Learning Approach

Camera and LiDAR Fusion for On-road Vehicle Tracking with Reinforcement Learning

Autonomous Multi-Sensor Fusion Techniques for Environmental Perception in Self-Driving Vehicles

Sensor Fusion: Gated Recurrent Fusion to Learn Driving Behavior from Temporal Multimodal Data

Radar and Camera Fusion for Multi-Task Sensing in Autonomous Driving

Real-Time Hybrid Multi-Sensor Fusion Framework for Perception in Autonomous Vehicles

Autonomous Navigation by Mobile Robot with Sensor Fusion Based on Deep Reinforcement Learning

Double Deep Q-Learning and Faster R-CNN-Based Autonomous Vehicle Navigation and Obstacle Avoidance in Dynamic Environment

Multi-objective Optimization Based Deep Reinforcement Learning for Autonomous Driving Policy

Robust Cognitive Capability in Autonomous Driving Using Sensor Fusion Techniques: A Survey

Deep Reinforcement Learning for Autonomous Driving with an Auxiliary Actor Discriminator

Multi-Modal Sensor Fusion-Based Deep Neural Network for End-to-End Autonomous Driving With Scene Understanding

Deep Reinforcement Learning framework for Autonomous Driving

Enabling Efficient Deep Convolutional Neural Network-based Sensor Fusion for Autonomous Driving

Autonomous Highway Driving using Deep Reinforcement Learning

Sensor and Sensor Fusion Technology in Autonomous Vehicles: A Review