Autonomous Vehicle Perception Fusion for Adverse-Weather Obstacle Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current vehicle perception systems face challenges in accurately detecting obstacles in adverse weather conditions and rely heavily on training materials, limiting their effectiveness in various environments.

Innovation Solution

A method and apparatus utilizing a perception detection network that combines sensor data from cameras and radars to enhance obstacle detection, incorporating 4D reconstruction and spatiotemporal analysis, enabling improved perception and control of vehicle driving through voxel-based obstacle representation and integration of navigation and traffic data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a pure vision manner is used for obstacle detection, then the system can identify obstacles after training and learning, but it has high dependency on training materials and limited effectiveness in various environments

Engineering Contradiction:
Improveobstacle detection accuracyVSAvoidenvironmental adaptability
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent combines multiple sensing modalities (vision sensors, radar, lidar) into a unified perception system. The vision system processes image data while radar/lidar provide depth and spatial information, creating a multi-sensor fusion architecture that overcomes the limitations of pure vision systems and improves both accuracy and environmental adaptability

Inventive Principle:
Principle #5Merging (Combining)

2Measurement precision

If lidar or millimeter-wave radar is used for obstacle detection, then the system can detect obstacles in surroundings, but detection accuracy is low in rainy or snowy weather

Engineering Contradiction:
Improveobstacle detection accuracyVSAvoidweather interference
Core Design Contradiction:
Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent creates a composite sensing system that integrates different sensor types (vision cameras, radar, lidar) with complementary characteristics. The vision system is less affected by weather compared to radar/lidar, and by fusing data from all sensors, the system compensates for individual sensor weaknesses in adverse weather conditions

Inventive Principle:
Principle #40Composite materials

3Reliability

If traditional perception systems are used, then the system can detect obstacles, but collision avoidance capability is limited

Engineering Contradiction:
Improvecollision avoidance reliabilityVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent introduces voxel-based 3D spatial representation and temporal dimension through 4D reconstruction (3D space + time). This multi-dimensional approach enables more accurate obstacle localization, trajectory prediction, and collision risk assessment, significantly improving collision avoidance capability while managing system complexity through structured data representation

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Data Source

PatentEP4714766A1Intelligent driving method and apparatus
Publication Date: 2026.03.25 YINWANG INTELLIGENT TECHNOLOGIES CO LTD
  • EP4714766A1 patent drawingFigure 1
  • EP4714766A1 patent drawingFigure 2
  • EP4714766A1 patent drawingFigure 3

AI summary

An intelligent driving method and an apparatus are disclosed. The method includes: obtaining collected data of a sensor of a vehicle for a scene, where the sensor includes at least one of a camera and a radar; inputting the collected data into a perception detection network, and outputting perception information, where the perception information indicates a voxel of an obstacle in a first scene; and controlling driving of the vehicle based on at least the perception information. This can enhance a perception capability of the vehicle for surroundings, help improve obstacle detection accuracy, and avoid a collision.