Platooning Vehicle Feature Map Sharing for Object Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for platooning vehicles face challenges in efficiently exchanging data between vehicles, leading to potential errors in object identification and increased resource consumption due to large data sizes and time delays in information transfer.
Innovation Solution
An electronic device equipped with a camera, communication circuitry, and a neural network model processes images to identify objects and transmit feature maps, allowing vehicles to independently verify classifications and efficiently share data, reducing resource usage and enhancing accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the leading vehicle transmits complete image data to following vehicles, then following vehicles can accurately identify objects, but data transmission size increases and consumes more resources
Solution Approach 1:
The patent extracts only the essential feature map data from the complete image, transmitting only this processed information to following vehicles. This allows following vehicles to perform object identification without receiving the entire image data, thereby reducing transmission size while maintaining identification accuracy.
Solution Approach 2:
The leading vehicle performs preliminary image processing to generate feature maps before transmission. By pre-processing the images to extract relevant features, the system reduces the data burden on following vehicles while ensuring they receive sufficient information for accurate object identification.
2Device complexity
If the leading vehicle processes all image data centrally, then object identification is simplified, but processing load and time delays increase
Solution Approach 1:
The patent segments the object identification process by having the leading vehicle generate feature maps while following vehicles perform the final classification. This distribution of processing tasks reduces the burden on any single vehicle and enables parallel processing, thereby reducing overall processing time.
Solution Approach 2:
The feature map serves as an intermediary data structure between the leading vehicle's image capture and the following vehicles' object identification. This intermediate representation facilitates efficient data transmission and enables following vehicles to independently complete the identification process without requiring extensive central processing.
3Reliability
If following vehicles receive and process large amounts of image data, then object recognition accuracy improves, but resource consumption increases
Solution Approach 1:
The patent extracts only the necessary feature map information from complete images, transmitting this condensed data to following vehicles. This approach maintains recognition reliability by preserving essential object features while significantly reducing the computational resources required for processing.
Solution Approach 2:
The leading vehicle performs preliminary feature extraction and generates optimized feature maps before transmission. This pre-processing step ensures that following vehicles receive data that is already optimized for recognition tasks, reducing their computational burden while maintaining high recognition reliability.
Data Source
AI summary
An electronic device may obtain an image through a camera. The electronic device may identify, using the image and a neural network model, a first classification result of an object included in the image. The electronic device may transmit, through communication circuitry, to external electronic devices included in vehicles subsequent to a vehicle equipped with the electronic device, a feature map of the neural network model based on the image. The electronic device may obtain, through the communication circuitry, from the external electronic devices, second classification results of the object which are calculated based on the feature map by the external electronic devices. The electronic device may determine, based on the first classification result and the second classification results, whether to drive the vehicle as a preceding vehicle.