Vehicle Discrepancy Distribution via Street Shadow Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern vehicles face challenges in accurately updating environmental information in real-time due to dynamic and unpredictable environmental conditions, leading to discrepancies between real-world conditions and the information used by vehicle systems, which can affect the proper functioning of vehicle functions such as navigation and driver-assisted controls.
Innovation Solution
A cloud-based vehicle information system generates and maintains 'street shadows' by collecting and propagating sensor data from vehicles and roadside resources, using machine learning to classify events and objects as short-lived or long-lived, and dynamically updating models to ensure accurate and timely information dissemination to vehicles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If periodic updates from a central server are used to update environmental conditions, then system complexity is reduced and ease of operation is improved, but the speed of information update deteriorates and reliability worsens due to insufficiently slow updates that cannot account for short-lived environmental conditions
Solution Approach 1:
The system segments the update mechanism into two parts: periodic central server updates for long-lived environmental conditions, and event-driven peer-to-peer propagation for short-lived conditions. This segmentation allows each part to specialize - the server handles comprehensive periodic updates while individual vehicles immediately share urgent changes, resolving the contradiction between update speed and system complexity.
Solution Approach 2:
The system performs preliminary classification of environmental conditions as either short-lived or long-lived. This preliminary action enables the system to pre-establish different update pathways: short-lived conditions trigger immediate peer-to-peer propagation while long-lived conditions wait for periodic server updates, thus achieving fast updates without requiring complex real-time synchronization across all vehicles.
2Reliability
If environmental conditions are updated in real-time through peer-to-peer vehicle communication, then reliability and measurement precision are improved, but device complexity and loss of time increase due to the need for discrepancy detection and propagation infrastructure
Solution Approach 1:
The system applies partial action by not requiring all vehicles to continuously monitor and verify all environmental conditions. Instead, when one vehicle detects a discrepancy for short-lived conditions, it immediately propagates this partial information to peers without requiring comprehensive verification from all vehicles, thus achieving high reliability with minimal time loss.
Solution Approach 2:
Each vehicle autonomously monitors its own sensor data against received environmental information and self-identifies discrepancies. This self-service approach eliminates the need for a centralized arbitration mechanism, reducing the time required for discrepancy detection while maintaining reliability through distributed verification across the vehicle network.
3Measurement precision
If machine learning classification is used to distinguish short-lived from long-lived events, then productivity and measurement precision are improved, but device complexity increases due to the need for machine learning models and classification infrastructure
Solution Approach 1:
The system introduces a lightweight intermediary classification layer that sits between raw sensor data and the update propagation mechanism. This intermediary uses simple machine learning models to classify events as short-lived or long-lived, providing measurement precision without requiring complex AI infrastructure in each vehicle - the classification service can be implemented as a centralized or edge-based service that vehicles query.
Data Source
AI summary
A vehicle information system and an in-vehicle agent implement a system for detecting discrepancies, for example due to drift, between actual real-world conditions in an environment surrounding a vehicle and corresponding conditions as represented in a model used by the vehicle. Additionally, the in-vehicle agent provides the detected discrepancies to a road-side resource and/or cloud-based computing device in order to distribute the detected discrepancies to other vehicles in the area. Also, a model maintained at the cloud-based computing devices may be updated to account for the discrepancies. The roadside resource and/or vehicle may infer estimated lifetimes of the detected discrepancies and intelligently distribute information describing the discrepancies based on their respective estimated lifetimes.