LiDAR Point Cloud Anomaly Detection Using Position-Based Scoring

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

LiDAR systems are vulnerable to attacks that can cause false object recognition due to manipulated laser light signals, posing safety risks in autonomous driving, and existing anomaly detection methods are computationally complex and inefficient.

Innovation Solution

A method for detecting anomalies in LiDAR point cloud data by determining anomaly scores based on position information using azimuth and vertical angles, with a detection module to identify tampered points, reducing computational complexity and enhancing security.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If existing anomaly detection methods are used to detect LiDAR point cloud data, then detection capability is provided, but computational complexity is high and efficiency is low

Engineering Contradiction:
Improveanomaly detection capabilityVSAvoidcomputational complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments the LiDAR point cloud data processing into distinct modules: position information extraction, specification-based expected position calculation, and anomaly scoring. Each module handles a specific aspect of the detection process, reducing overall computational complexity while maintaining detection capability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts only the essential position information (azimuth angle and vertical angle) from the LiDAR point cloud data, rather than processing all raw data. This selective extraction reduces computational load while preserving the key features needed for anomaly detection.

Inventive Principle:
Principle #2Taking out (Extraction)

2Measurement precision

If comprehensive LiDAR specifications are used for accurate anomaly detection, then detection precision is improved, but computational overhead increases

Engineering Contradiction:
Improveanomaly detection precisionVSAvoidcomputational overhead
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent applies local quality by using specific LiDAR specifications (horizontal angular resolution, vertical angle per channel, channel count) that are directly relevant to position calculation, rather than processing all possible sensor parameters. This selective use of specifications maintains precision while reducing computational overhead.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent transforms the anomaly detection problem into a parameter comparison task, where the actual position parameters are compared against expected position parameters calculated from LiDAR specifications. This parameter-based approach simplifies the detection process while maintaining accuracy.

Inventive Principle:
Principle #35Parameter changes

3Reliability

If real-time anomaly detection is implemented in autonomous driving systems, then safety is improved, but processing speed requirements increase system complexity

Engineering Contradiction:
Improveautonomous driving safetyVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent performs preliminary calculations of expected position information based on LiDAR specifications before actual anomaly detection. By pre-establishing the relationship between sensor specifications and expected data positions, the system reduces real-time computational requirements and simplifies the overall system architecture.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4363888B1Method for detecting anomalies of lidar point cloud data and related device
Publication Date: 2026.04.15 YINWANG INTELLIGENT TECHNOLOGIES CO LTD
  • EP4363888B1 patent drawingFigure 1~2
  • EP4363888B1 patent drawingFigure 3~4
  • EP4363888B1 patent drawingFigure 5

AI summary

Embodiments of this application provide a method for detecting anomalies of LiDAR point cloud data and related device, the method including: obtaining a first position information, wherein the first position information indicating a position of a target data point in the LiDAR point cloud data; determining a second position information corresponding to the target data point according specifications of a LiDAR sensor; determining the target data point is an anomaly point or a normal point according to the first position information and the second position information. The above-mentioned technical solution can detect the anomaly point efficiently.