Radar-based Target Set Generation

Simple SummaryContent extracted from patent full text and abstract with AI.

This patent describes a method and system for accurately detecting and locating multiple targets (such as vehicles or pedestrians) using radar, particularly millimeter-wave radar. The core innovation is using deep neural networks, specifically a convolutional encoder followed by fully-connected layers, to process radar images and output the coordinates of detected targets. The process includes advanced data transformations, such as non-uniform discrete Fourier transforms, to improve detection accuracy and efficiency, and uses specially designed neural network training methods for optimal performance.

Use CasesContent extracted from patent full text and abstract with AI.

• Automotive driver assistance systems (ADAS) and collision avoidance in cars
• Autonomous vehicle navigation and obstacle detection
• Industrial automation and robotics safety systems
• Surveillance and security systems (e.g., monitoring restricted areas for intruders)
• Smart traffic management systems (tracking and counting vehicles or pedestrians)
• Drones or airborne vehicles for navigation and ground mapping
• Healthcare monitoring (e.g., tracking patient movement or falls)
• Search and rescue operations in dangerous or low-visibility environments

BenefitsContent extracted from patent full text and abstract with AI.

• Accurate real-time detection and localization of multiple moving or static targets
• Improved robustness in challenging environments with noise or low visibility
• Efficient processing through advanced neural network techniques, reducing required computational resources
• Ability to handle and distinguish between multiple targets simultaneously (multi-target tracking)
• Flexible architecture suitable for 2D and 3D target localization
• Learned data processing (e.g., non-uniform DFT) for tailored performance and memory savings
• Potential for integration in compact, low-cost radar devices due to efficient processing pipeline
• Enhanced reliability via confidence assessment for detected targets (e.g., using SNR)

Technical Classifications (CPCs)

Inventors & Applicants

Patent Abstract

In an embodiment, a method for generating a target set using a radar includes: generating, using the radar, a plurality of radar images; receiving the plurality of radar images with a convolutional encoder; and generating the target set using a plurality of fully-connected layers based on an output of the convolutional encoder, where each target of the target set has associated first and second coordinates.