Vehicle Object Tracking Box Merging for LiDAR Route Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In vehicle control systems using LiDAR sensors, incorrect merging of virtual boxes can lead to misidentification of external objects' type, state, or size, potentially causing dramatic or incorrect changes in the driving route.
Innovation Solution
A vehicle control apparatus and method that utilize a processor to generate a tracking box from a virtual box, determine associated virtual boxes, and merge them based on distance, object separation from the road edge, and object type, while controlling the merging process based on the size of the merge box and its difference from the determined virtual boxes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If virtual boxes are merged based on LiDAR data, then object identification is achieved, but incorrect merging causes misidentification of object type, state, or size
Solution Approach 1:
The system performs preliminary actions by generating virtual boxes at a first time point, converting them to tracking boxes, and determining associated virtual boxes at a second time point before merging. This preliminary processing ensures that only properly associated virtual boxes are merged, preventing incorrect object identification and maintaining driving route correctness.
Solution Approach 2:
The system uses feedback by determining whether virtual boxes are associated with the tracking box based on overlap ratios and correlation distances, then using this information to control the merging process. This feedback mechanism ensures that merging only occurs when virtual boxes truly represent the same object, improving identification accuracy while preventing erroneous merges that would compromise reliability.
2Measurement precision
If all virtual boxes are merged, then complete object detection is achieved, but processor load increases significantly
Solution Approach 1:
The system applies partial action by determining and merging only the virtual boxes that are associated with the tracking box, rather than merging all virtual boxes. This is controlled through overlap ratio thresholds and correlation distance criteria, ensuring complete object detection while avoiding unnecessary processing of unrelated virtual boxes, thus maintaining processor efficiency.
3Productivity
If virtual boxes are merged based on distance alone, then processing speed is maintained, but object identification accuracy decreases
Solution Approach 1:
The system changes the parameters used for merging by incorporating multiple criteria beyond just distance: overlap ratio between tracking box and virtual boxes, correlation distance between points, object separation from road edges, and object type information. This multi-parameter approach maintains processing speed through efficient calculations while significantly improving object identification accuracy compared to distance-only merging.
Data Source
AI summary
An apparatus for vehicle control is introduced. The apparatus includes a sensor and a processor configured to generate a tracking box based on a virtual box obtained from the sensor at a first time point, converting it to a second time point, and determine virtual boxes associated with the tracking box from a set of virtual boxes at the second time point, derived from sensor data. The processor decides whether to merge some or all of these virtual boxes based on their inter-distances, whether any virtual box corresponds to an object separated from a road edge, or if they are contained within the tracking box. The merging process is controlled based on the resulting merge box size, the difference from the size based on the identified virtual boxes, or the object type. The apparatus outputs a signal to indicate the outcome of this controlled merging.