Vehicle Platoon Scene Perception Using Distributed Neural Predictions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In vehicle platoons, following vehicles do not contribute significantly to the driving performance of the lead vehicle, as they primarily focus on maintaining distance and lane position, leading to underutilization of resources such as sensors and processing power.
Innovation Solution
Implementing a system where following vehicles in a platoon process raw data from the lead vehicle using their own neural networks, which are different from the lead vehicle's, and combine their predictions with the lead vehicle's to enhance scene perception and accuracy, thereby redirecting saved processing power to improve the overall platoon's performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If following vehicles focus on maintaining distance and lane position, then vehicle platoon stability is improved, but resource utilization of following vehicles deteriorates
Solution Approach 1:
Following vehicles are assigned multiple functions: they maintain platoon stability through distance and lane position control, while simultaneously contributing to scene perception by processing raw data through their neural networks. This multi-functionality transforms underutilized computational resources into valuable assets for collective intelligence, allowing following vehicles to serve both stability maintenance and perception enhancement roles.
2Measurement precision
If following vehicles process raw data with their own neural networks, then scene perception accuracy is improved, but system complexity increases
Solution Approach 1:
The scene perception task is segmented and distributed across multiple vehicles in the platoon. Each following vehicle processes a portion of the raw data through its own neural network independently, then contributes its predictions to the collective result. This segmentation reduces the computational burden on any single vehicle while improving overall perception accuracy through diversified processing perspectives.
Solution Approach 2:
The system merges predictions from multiple neural networks located in different vehicles (lead vehicle and following vehicles) to generate a consolidated scene perception result. By combining the outputs of these distributed neural networks, the system achieves higher accuracy than any single vehicle could attain alone, while the modular architecture manages complexity through standardized communication protocols.
3Productivity
If following vehicles contribute to driving performance, then overall platoon performance is improved, but communication load increases
Solution Approach 1:
The system extracts only the essential prediction results from following vehicles and transmits them to the lead vehicle, rather than sharing all raw data or intermediate processing outputs. This extraction approach minimizes communication load by transmitting only the valuable processed information needed for improved driving performance, filtering out redundant data transmissions.
Data Source
AI summary
A vehicle includes one or more sensors configured to obtain raw data related to a scene, one or more processors, and machine readable instructions stored in one or more memory modules. The one machine readable instructions, when executed by the one or more processors, cause the vehicle to: process the raw data with a first neural network stored in the one or more memory modules to obtain a first prediction about the scene, transmit the raw data to a computing device external to the vehicle, receive a second prediction about the scene from the computing device in response to transmitting the raw data to the computing device, and determine an updated prediction about the scene based on a combination of the first prediction and the second prediction.