Neural Network Electronic Road Map Lane Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High definition maps required for autonomous vehicles are expensive to produce and consume significant computational resources, limiting their availability for all locations.
Innovation Solution
A neural network system that configures an electronic road map to distinguish lanes by detecting features in images and correlating them with prior feature maps, producing a labeled training map to improve lane recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high definition maps are produced using traditional methods, then lane recognition accuracy is improved, but production cost and computational resource consumption increase significantly
Solution Approach 1:
The system performs preliminary actions by collecting and storing road feature data (lane markings, road signs, geometry) in advance to create feature maps. These pre-collected features are then used during neural network training to improve lane recognition accuracy without requiring complex real-time processing, thus reducing computational resource consumption during operation.
Solution Approach 2:
The system creates simplified copies of road features by extracting key characteristics (lanes, curbs, signs) from images and representing them as structured feature data. These feature copies are stored in feature maps that can be efficiently processed during training, avoiding the need to process full-resolution images repeatedly, thereby reducing computational complexity while maintaining recognition accuracy.
2Measurement precision
If high definition maps are produced using traditional methods, then lane recognition accuracy is improved, but production cost increases
Solution Approach 1:
The system creates simplified digital copies of road features (lane markings, curbs, signs) from images and stores them as structured data in feature maps. This copying approach enables efficient neural network training without requiring expensive manual annotation of complete high-definition maps, significantly reducing production costs while maintaining lane recognition accuracy.
Solution Approach 2:
The system performs preliminary feature extraction and organization during data collection phases, creating ready-to-use feature maps before training. This preliminary action reduces the need for expensive post-processing and manual intervention, lowering overall production costs while ensuring high lane recognition accuracy through pre-organized training data.
3Use of energy by moving object
If traditional electronic road maps are used, then computational resources are saved, but accuracy for autonomous vehicle navigation is insufficient
Solution Approach 1:
The system segments road information into distinct feature categories (lanes, curbs, signs, geometry) and organizes them in structured feature maps. This segmentation allows the neural network to process only relevant features during training and operation, reducing computational resource consumption compared to processing complete high-definition maps, while improving navigation accuracy through focused feature analysis.
Solution Approach 2:
The system extracts only the essential road features (lane markings, curbs, signs, road geometry) from images and stores them in feature maps, discarding unnecessary image data. This extraction approach maintains navigation accuracy by preserving critical features while significantly reducing computational resource requirements compared to using traditional complete high-definition maps.
Data Source
AI summary
A neural network can be configured to produce an electronic road map. The electronic road map can have information to distinguish lanes of a road. A feature in an image can be detected. The image can have been produced at a current time. The image can be of the road. The feature in the image can be determined to correspond to a feature, of a plurality of features, in a feature map. The feature map can have been produced at a prior time from one or more images. A labeled training map can be produced from the feature in the image and the plurality of features in the feature map. The labeled training map can have the information to distinguish the lanes of the road. The neural network can be trained to produce, in response to a receipt of the image and the feature map, the labeled training map.