Radar Data Processing for Dense Point Cloud Generation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Radar sensors struggle to achieve artificial intelligence performance comparable to LiDAR sensors due to sparse point cloud data, limiting object recognition and environmental awareness in applications like autonomous driving.

Innovation Solution

A system and method that processes radar data by acquiring and converting radar data from multiple time points to generate refined data, incorporating time-difference-based trajectory processing, and training an artificial intelligence model using combined radar and LiDAR data to enhance object recognition.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If radar data from multiple time points is converted and combined for trajectory processing, then point cloud density and object recognition accuracy are improved, but data processing complexity and computational load increase

Engineering Contradiction:
Improveobject recognition accuracyVSAvoiddata processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs preliminary conversion of historical radar data to predicted present positions before combining with current data. This advance preparation of trajectory information allows the AI model to receive pre-processed, density-enhanced point clouds, improving recognition accuracy while managing complexity through structured preprocessing steps

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent implements nested processing where multiple levels of data refinement are applied: raw radar data is converted to predicted positions, then combined with current data, and finally processed through trajectory processing algorithms. This nested structure systematically enhances point cloud density while organizing computational complexity into manageable hierarchical layers

Inventive Principle:
Principle #7Nested doll (Nesting)

2Reliability

If multiple frames of radar data are accumulated and processed for trajectory enhancement, then artificial intelligence performance is improved, but processing time and computational resources increase

Engineering Contradiction:
Improveartificial intelligence performanceVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

Historical radar data is converted to predicted present positions in advance, before being combined with current frame data. This preliminary conversion of multiple frames prepares trajectory information ahead of time, enabling the AI model to utilize enhanced data without incurring excessive processing delays during critical detection phases

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system maintains continuous accumulation and processing of radar data frames, creating an ongoing trajectory enhancement process. This continuous operation ensures that the AI model consistently receives high-quality, density-enhanced point clouds, improving overall system reliability and AI performance through sustained data quality

Inventive Principle:
Principle #20Continuity of useful action

Data Source

PatentEP4575574A1System and method for processing radar data
Publication Date: 2025.06.25 BITSENSING INC
  • EP4575574A1 patent drawingFigure 1~2
  • EP4575574A1 patent drawingFigure 3A~3B
  • EP4575574A1 patent drawingFigure 4

AI summary

A system for processing radar data includes a radar device, at least one processor, and at least one memory including a computer program code. Wherein the at least one memory and the computer program code are configured to, with the at least one processor, cause the system to, acquire, from the radar device, first radar data corresponding to a target area and second radar data generated before the first radar data, convert at least a part of the second radar data based on a time difference between a time point at which the first radar data is acquired and a time point at which the second radar data is acquired, and generate refined data by considering the converted second radar data for trajectory processing along with the first radar data.