Vehicle Environment Recognition with Adaptive Point Cloud Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing external environment recognition systems for vehicles face high processing loads due to increased detection points, leading to inefficiencies in data processing and potential hardware cost increases, especially in complex environments.
Innovation Solution
The system intermittently irradiates electromagnetic waves in a discrete manner, adjusting detection point density based on distance from the vehicle, reducing the number of detection points required for recognition processing while maintaining accuracy, and optimizing hardware resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If scanning irradiation is performed in all directions within the field of view to acquire comprehensive point cloud data, then recognition accuracy of the external environment is improved, but the number of detection points increases leading to higher processing load
Solution Approach 1:
The patent applies local quality by differentiating detection point density based on spatial location and object type. High-density detection points are allocated to critical areas such as the road surface in the traveling direction and regions containing three-dimensional objects, while lower density is used in less critical areas. This selective allocation maintains recognition accuracy for important elements while reducing the total number of detection points to lower processing load.
Solution Approach 2:
The patent segments the field of view into multiple regions with different detection requirements. The road surface area in the traveling direction is separated from other areas, and three-dimensional object regions are identified distinctively. Each segment receives appropriate detection point density according to its specific recognition needs, optimizing the balance between accuracy and processing load.
2Measurement precision
If the number of detection points is increased to improve recognition accuracy in complex environments, then measurement precision is improved, but hardware costs and processing complexity increase
Solution Approach 1:
The patent implements partial action by acquiring detection points selectively rather than uniformly across the entire field of view. Detection points are concentrated in partial regions where they are most needed (road surface in traveling direction, three-dimensional object locations) while omitting or reducing points in less critical areas. This approach achieves sufficient recognition accuracy without the hardware and processing overhead of complete coverage.
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 processing load and hardware costs without compromising recognition accuracy, enabling efficient and cost-effective external environment recognition for vehicles.
Implementation Method 1
acquire, frame by frame, point cloud data including three-dimensional position information of detection points on surfaces of objects around a subject vehicle based on reflected waves from the objects
Data Source
AI summary
The external environment recognition apparatus includes an in-vehicle detector and a microprocessor. The microprocessor recognizes, as road surface information, a road surface of a road on which a subject vehicle travels and a three-dimensional object thereon, based on point cloud data of each frame acquired by the in-vehicle detector; determines, based on a predetermined size of a predetermined three-dimensional object and a measurement distance from the subject vehicle to the object based on the point cloud data, an interval of detection points for point cloud data of a next frame; and predicts a gradient of an unrecognized road surface based on gradient information associated with map information. The microprocessor determines, for a range from the maximum depth distance to the required depth distance, the interval based on the predetermined size and an estimated distance to the object estimated from the map information and the gradient.