MMW Radar-Based Technologies in Autonomous Driving: A Review.

Area of Science:

• Robotics and Intelligent Systems
• Sensor Fusion for Autonomous Systems
• Machine Learning for Environmental Perception

Background:

• Automated vehicles (AVs) require advanced environmental perception.
• Millimeter-wave (MMW) radar is a key sensor due to its cost-effectiveness, all-weather adaptability, and motion detection capabilities.
• Existing research lacks a comprehensive survey on deep learning applied to MMW radar data for AVs.

Purpose of the Study:

• To provide an overview of state-of-the-art radar-based technologies in AVs.
• To survey deep learning applications utilizing MMW radar data for autonomous driving.
• To identify current challenges and future research directions.

Main Methods:

• Introduction to MMW radar data models and representations.
• Review of radar-based applications in AVs, including Advanced Driving-Assistance Systems (ADAS).
• Analysis of deep learning techniques for object detection, tracking, motion prediction, and self-localization using radar data.

Main Results:

• MMW radar data offers diverse types for various autonomous driving levels.
• Radar is crucial for ADAS and high-level autonomous functions like object detection and localization.
• Deep learning models are increasingly vital for extracting complex information from radar data.

Conclusions:

• Deep learning applied to MMW radar data is essential for robust AV perception.
• Further research is needed to address challenges in data representation and model generalization.
• Future work should focus on enhancing sensor fusion and real-time processing for safety-critical applications.