LiDAR Detection Point Density Control for Processing Load Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing external environment recognition systems for vehicles face high processing loads due to numerous detection points, especially in scenarios with many objects on the road, leading to increased data capacity and potential recognition accuracy degradation.
Innovation Solution
The system intermittently irradiates LiDAR light in the advancing direction of the vehicle, setting higher detection point density on the road surface far from the vehicle and lower closer to it, reducing the total number of detection points for recognition processing without compromising accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the LiDAR scans and emits electromagnetic waves to detect the external environment, then recognition accuracy is improved, but the number of detection points increases leading to higher processing load
Solution Approach 1:
The patent applies local quality by differentiating the density of detection points based on their distance from the vehicle. Detection points on the road surface far from the vehicle are set with higher density, while those closer to the vehicle have lower density. This localized variation in detection quality optimizes the balance between recognition accuracy for distant objects and processing load reduction.
Solution Approach 2:
The patent segments the detection space into different regions based on distance from the vehicle. By dividing the detection area into near-field and far-field zones, the system can apply different detection point densities to each segment, reducing the overall number of detection points while maintaining accuracy where needed.
2Measurement precision
If the scanning angular resolution is increased to detect smaller objects, then recognition accuracy is improved, but the number of detection points increases exceeding the limit value
Solution Approach 1:
The patent applies local quality by differentiating the density of detection points based on their distance from the vehicle. Detection points on the road surface far from the vehicle are set with higher density, while those closer to the vehicle have lower density. This localized variation in detection quality optimizes the balance between recognition accuracy for distant objects and processing load reduction.
Solution Approach 2:
The patent dynamically adjusts the scanning angular resolution and detection point density based on the distance to detected objects. When objects are detected at greater distances, the system increases angular resolution and detection density in those regions, while reducing it for nearer objects, thereby adapting the detection parameters to the actual recognition needs.
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 reduces the number of detection points required for recognition processing, thereby lowering processing loads while maintaining recognition accuracy, even in complex environments.
Implementation Method 1
a device for respectively changing the irradiation angles of laser light irradiated from a LiDAR
Implementation Method 2
an in-vehicle detector configured to scan and emit an electromagnetic wave in a first direction and as a second direction intersecting the first direction to detect an external environment situation
Data Source
AI summary
An external environment recognition apparatus includes an in-vehicle detector and a microprocessor. The microprocessor is configured to perform: recognizing a surface of the road and a three-dimensional object on the road for each frame as the road surface information based on a three-dimensional point cloud data including distance information for every one of a plurality of detection points in a matrix shape acquired in every frame by the in-vehicle detector; determining an interval of detection points of the three-dimensional point cloud data as a scanning angular resolution of the electromagnetic wave based on a size of a predetermined three-dimensional object and a distance to the predetermined three-dimensional object; and taking a predetermined avoidance measure so that at least one of the scanning angular resolution and a number of detection points corresponding to the interval of detection points does not exceed a predetermined limit value.