Incremental Neural Network Updates for Autonomous Vehicle Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for autonomous vehicles lack efficient mechanisms for incrementally updating trained neural networks based on data from edge computing devices and infrastructure, limiting their ability to adapt to changing environments and improve route planning accuracy.
Innovation Solution
A system that uses a processor to determine vehicle routes via a trained neural network model, updates the model based on data from edge computing devices and infrastructure, and incorporates data from other vehicles, allowing for over-the-air updates and weight adjustments, which are then uploaded to a specification platform for further refinement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a trained neural network model is used for route determination in autonomous vehicles, then route planning capability is achieved, but the system cannot adapt to changing environments and improve accuracy over time
Solution Approach 1:
The neural network model is transformed from a static, pre-trained system to a dynamic, continuously learning system. The model receives incremental updates through edge computing devices and infrastructure, allowing it to adapt its weights and structure based on real-time environmental data and changing conditions, thereby simultaneously improving adaptability and maintaining reliability through continuous refinement
Solution Approach 2:
A feedback loop is established where the neural network model's performance is continuously monitored through data collected by edge computing devices and infrastructure. This feedback mechanism enables the system to identify areas for improvement and automatically adjust the model through incremental learning, ensuring both adaptability to new conditions and sustained route planning accuracy
2Adaptability or versatility
If the neural network model is updated frequently to improve adaptability, then responsiveness to environmental changes improves, but computational overhead and system complexity increase
Solution Approach 1:
The model update process is segmented into distributed edge computing nodes and centralized infrastructure components. Each edge device independently processes local data and performs localized model updates, while the centralized infrastructure coordinates overall model management. This segmentation reduces the computational burden on any single component and simplifies the overall system architecture while maintaining high responsiveness
Solution Approach 2:
Edge computing devices serve as intermediaries between the autonomous vehicle's neural network model and the centralized infrastructure. These intermediaries buffer and manage the complexity of frequent updates by pre-processing data locally and selectively transmitting only essential update information to the infrastructure, thereby reducing system complexity while maintaining adaptability
Data Source
AI summary
A system comprises a computer including a processor and a memory. The memory includes instructions such that the processor is programmed to: determine, via a trained neural network model, a route for a vehicle to traverse based on vehicle sensor data, and update the trained neural network model based on data received from at least one of an edge computing device or an infrastructure (V2I) device.