Stationary Object Detection Using Triangular Network Feature Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting stationary objects from point cloud data, such as bounding box detection, often result in missed or false detections due to inaccuracies in sizing and orientation, leading to poor accuracy in static maps, which is crucial for safe driving in driverless vehicles.
Innovation Solution
A method involving obtaining point cloud data and feature information, generating a triangular network model, and using a classification model to accurately identify stationary objects by inputting feature maps, including intensity, planar, and depth information, to distinguish between background and stationary objects, with optional pre-processing to remove moving objects and camera parameter optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If bounding box detecting method is used to identify stationary vehicles, then the detection process is simple, but the accuracy is poor due to missed detection or misdetection
Solution Approach 1:
The patent segments the detection process into multiple stages: point cloud data acquisition, feature extraction (intensity, geometric, contextual features), candidate region generation, and classification. This multi-stage segmentation allows each stage to focus on specific aspects, improving overall detection accuracy while maintaining manageable complexity
Solution Approach 2:
The patent introduces feature maps as intermediary representations between the raw point cloud data and the final detection results. These feature maps encode multiple types of information (intensity, geometry, context) and serve as a bridge that enables more accurate classification while keeping the overall system structure organized
2Measurement precision
If feature map with multiple feature types is used for classification, then the detection accuracy is improved, but the computational complexity increases
Solution Approach 1:
The patent performs preliminary organization of point cloud data into structured formats and pre-computes feature statistics before the main classification process. This preliminary action reduces the computational burden during classification by having data ready in an optimized format
Solution Approach 2:
The patent employs dynamic feature selection and adaptive processing where not all feature types are processed with equal depth for every data point. The system adapts the level of processing based on local characteristics, optimizing the balance between accuracy and computational cost
Data Source
AI summary
A stationary object detecting method, a stationary object apparatus, and an electronic device are disclosed in embodiments of the present disclosure, the method includes: obtaining point cloud data of a scene and feature information of each of data points in the point cloud data; performing a triangulation network connection on each of the data points in the point cloud data to generate a triangular network model, and taking a picture of the triangular network model using a preset camera to obtain an image; obtaining a first data point corresponding to each pixel point in the image, and obtaining a feature map of the point cloud data according to feature information of each first data point; inputting the feature map into a classification model to obtain data points corresponding to the stationary object in the point cloud data.