LiDAR Object Detection Using Map Feature Fusion for Autonomous Driving

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Autonomous driving vehicles (ADVs) face reduced precision in object detection when relying solely on sensor systems without incorporating high definition map information, leading to potential safety and reliability issues in navigation.

Innovation Solution

Combining map features extracted by a convolution neural network with point cloud features from LiDAR data using a feature learning network, and feeding the combined features to neural networks for enhanced object detection, thereby integrating environmental map data into the ADV's perception system.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If ADV relies solely on sensor systems for object detection, then device complexity is reduced, but measurement precision deteriorates

Engineering Contradiction:

Improvedetection system complexityVSAvoidobject detection precision

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent combines map data from high definition maps with real-time sensor data from LiDAR and other sensors to create a fused detection result. This merging of multiple data sources improves object detection precision by compensating for the limitations of individual sensor systems while maintaining a manageable system architecture through modular integration.

Inventive Principle:

Principle #5Merging (Combining)

2Measurement precision

If ADV incorporates high definition map information for object detection, then measurement precision is improved, but device complexity increases

Engineering Contradiction:

Improveobject detection precisionVSAvoiddetection system complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the object detection process into distinct modules: map data processing, sensor data processing, and fused detection. This segmentation allows each module to be optimized independently and processed in a structured manner, managing system complexity through modular architecture while achieving improved detection precision through comprehensive data fusion.

Inventive Principle:

Principle #1Segmentation

3Loss of time

If ADV uses only sensor data without map information, then information processing time is reduced, but loss of information increases

Engineering Contradiction:

Improveinformation processing timeVSAvoidenvironmental context information

Core Design Contradiction:

Loss of timeVSLoss of information

Solution Approach 1:

The patent performs preliminary processing of map data to extract relevant environmental context information before fusion with sensor data. This preliminary action prepares the map information in advance, allowing it to be efficiently integrated with real-time sensor data without causing excessive processing delays, thus reducing information loss while managing processing time.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11520347B2Comprehensive and efficient method to incorporate map features for object detection with LiDAR

Publication Date: 2022.12.06 BAIDU USA LLC

AI summary

According to various embodiments, systems and methods described in the disclosure combine mapped features with point cloud features to improve object detection precision of an autonomous driving vehicle (ADV). The map features and the point cloud features can be extracted from a perception area of the ADV within a particular angle view at each driving cycle based on a position of the ADV. The map features and the point cloud features can be concatenated and provided to a neutral network for object detections.