Multi-LIDAR Object Recognition Using Segmented Contour Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing object recognition systems for autonomous vehicles face challenges in accurately identifying the position of objects, particularly in confined spaces like tunnels, due to the disparity in data from multiple LIDAR sensors and the need for reduced computational load.

Innovation Solution

The system employs a processor to identify contour points of stationary objects by processing data from multiple LIDAR sensors, adjusting candidate contour points based on sensor direction and distance, and generating signals to display object locations on a map, thereby enhancing accuracy and reducing computation when identifying specified objects.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If data from multiple LIDAR sensors is processed to increase measurement coverage, then object position identification accuracy is improved, but computational load increases

Engineering Contradiction:
Improveobject position identification accuracyVSAvoidcomputational load
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent segments the contour points into multiple groups, where each group corresponds to a specific LIDAR sensor. The processor then processes each group separately rather than handling all points uniformly, which reduces computational complexity while maintaining the benefits of multi-sensor data fusion for improved position accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different processing strategies to different groups of contour points based on their source sensors and spatial locations. By treating each sensor's data with appropriate local processing rules and weighting factors, the system optimizes computational efficiency for each region while achieving overall high accuracy through localized quality adjustments.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If contour points from multiple LIDAR sensors are integrated to improve location accuracy in confined spaces, then object recognition accuracy is improved, but data processing complexity increases

Engineering Contradiction:
Improvelocation accuracy in confined spacesVSAvoiddata processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the contour points into groups corresponding to different LIDAR sensors, enabling modular processing. Each group can be processed independently with specific algorithms tailored to that sensor's characteristics, reducing the overall complexity of integrating multi-sensor data while maintaining high location accuracy in confined spaces like tunnels.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary processing layer that handles the complexity of multi-sensor data fusion. This intermediary structure manages the coordination between different LIDAR sensors, resolves discrepancies in their data, and synthesizes accurate location information without exposing the complexity of the fusion process to the overall system.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Productivity

If all contour points are processed to ensure complete object identification, then recognition completeness is improved, but computational time increases

Engineering Contradiction:
Improverecognition completenessVSAvoidcomputational time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent applies partial processing by selectively processing contour point groups based on their importance and spatial distribution. Instead of uniformly processing all points, the system identifies and processes only the necessary portions of contour data required for complete object recognition, reducing computational time while maintaining recognition completeness through strategic sampling and prioritization.

Inventive Principle:
Principle #16Partial or excessive action

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach increases the accuracy of object position identification and reduces computational requirements by effectively integrating data from multiple LIDAR sensors and optimizing processing based on object type and location.

Implementation Method 1

The LIDAR may obtain a distance from the LIDAR to the object through an interval between a time point at which the laser is transmitted and a time point at which the laser reflected on the object is received

Methodology Applied
Scientific EffectTime of Flight: Time of Flight

Implementation Method 2

a time point at which the laser reflected on the object is received

Methodology Applied
Scientific EffectLight Reflection: Reflection

Data Source

PatentUS20250061683A1Apparatus For Recognizing Object And Method Thereof
Publication Date: 2025.02.20 HYUNDAI MOTOR CO LTD
  • US20250061683A1 patent drawing
  • US20250061683A1 patent drawing
  • US20250061683A1 patent drawing

AI summary

An apparatus may include a first LIDAR, a second LIDAR, and a processor. The processor may identify contour points of an object by using or changing points obtained through any one of the first LIDAR and the second LIDAR among points obtained through the first LIDAR and included in a specific frame and points obtained through the second LIDAR and included in the specific frame, based on obtaining a signal indicating a limited space where a host vehicle is located, from the map system. The points obtained through the first LIDAR and included in the specific frame may be obtained from a same object as the points obtained through the second LIDAR and included in the specific frame.