3D Object Detection Using Map-Guided Crossing Area Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional obstacle detection systems are unable to detect objects that are not on the surface of the planned travel route of a host vehicle and therefore cannot anticipate objects that may cross the vehicle in the future.
Innovation Solution
An object detection method and apparatus that utilize three-dimensional distance measurement data, combined with map data and vehicle information, to estimate and detect crossing object existence areas where objects are likely to intersect with the host vehicle's planned travel path, incorporating a central controller with functional units for object detection and tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If obstacle detection is performed only on the surface of the planned travel route, then the detection process is simple and fast, but objects that are not currently on the route surface cannot be detected
Solution Approach 1:
The patent performs preliminary detection in areas where objects are likely to appear before they actually enter the planned travel route. By expanding the detection area to include regions adjacent to and around the planned route, the system detects objects in advance of their actual path intersection, enabling proactive safety measures.
Solution Approach 2:
The patent transitions from two-dimensional route surface detection to three-dimensional spatial detection. By incorporating vertical height information and expanding detection to volumetric regions around the planned route, the system captures objects at various heights and positions that would be invisible to traditional surface-only detection.
2Reliability
If the detection area is expanded to cover larger surrounding regions, then more objects can be detected beforehand, but the data processing load increases
Solution Approach 1:
The patent applies different detection strategies to different spatial zones. High-priority processing is applied to areas immediately adjacent to the planned route where objects are most likely to intersect, while lower-priority or filtered processing is applied to more distant regions. This localized quality approach optimizes processing resources according to actual risk levels.
Solution Approach 2:
The detection area is segmented into multiple zones based on proximity to the planned route and likelihood of object intersection. The system processes data from different segments with appropriate priority levels, focusing computational resources on high-risk zones while reducing processing for low-risk areas, thereby managing overall data processing load efficiently.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
An object detection method according the present invention includes: acquiring three-dimensional data on an area around a host vehicle by use of a distance measurement sensor; based on map data on an area around a current position of the host vehicle, setting a planned travel area where the host vehicle is going to travel in the future; estimating crossing object existence areas where there currently exist objects which are likely to cross the host vehicle in the future in the set planned travel area; and detecting the object by use of the three dimensional data on insides of the estimated crossing object existence areas.