Autonomous Vehicle Perception via Edge Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous vehicle systems face challenges in determining optimal actions due to the limitations of existing sensor data processing and communication between onboard, edge, and cloud layers, which can lead to inefficiencies in decision-making and action execution.
Innovation Solution
A method that involves obtaining and transforming onboard vehicle sensor data and edge sensor data into a unified perception output, allowing for coordinated action determination and execution across vehicle, edge, and cloud layers, utilizing edge nodes for real-time processing and cloud layers for supplementary information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If onboard vehicle sensors alone are used for autonomous vehicle action determination, then the system complexity is reduced, but the measurement precision and reliability of environmental perception deteriorate
Solution Approach 1:
The patent combines onboard vehicle sensors with edge layer sensors to create a unified perception system. The edge layer sensors are integrated with the vehicle's sensor system, allowing data from both sources to be merged and processed together, thereby improving measurement precision without significantly increasing system complexity
Solution Approach 2:
The edge layer sensors serve multiple functions: they provide environmental perception data for autonomous vehicle operation, enable improved trajectory prediction, and support both vehicle-determined and edge-determined action modes. This multi-functionality improves perception precision while avoiding additional dedicated systems
2Reliability
If edge layer sensors are integrated with onboard vehicle sensors, then the reliability of autonomous vehicle action determination is improved, but the device complexity increases
Solution Approach 1:
The edge layer sensors are merged with the onboard vehicle sensors into a unified sensor system. This integration allows the system to process sensor data from multiple sources simultaneously, improving the reliability of action determination while managing complexity through unified processing architecture
Solution Approach 2:
The patent introduces an intermediary processing layer that handles data from both onboard and edge sensors. This intermediary layer coordinates the data flow and processing between the two sensor systems, improving reliability through combined data while preventing complexity escalation through structured mediation
3Ease of operation
If vehicle sensor data is transformed to homogenous coordinate system with edge sensor data, then the ease of operation for unified perception is improved, but the loss of time in data processing increases
Solution Approach 1:
The system performs preliminary coordinate transformations and data preprocessing at the edge layer before the vehicle receives the data. By pre-transforming edge sensor data to the homogenous coordinate system in advance, the vehicle can directly use the processed data without performing additional transformations, thus improving ease of operation while minimizing time loss
Solution Approach 2:
The edge layer autonomously performs data transformation and preprocessing operations on its own sensor data before transmission to the vehicle. This self-service approach allows the edge layer to prepare data in the required format independently, reducing the processing burden and time requirements for the vehicle system
Data Source
AI summary
An autonomous vehicle (AV) perception system and method of determining an autonomous vehicle (AV) action for a host vehicle. The method includes: obtaining onboard vehicle sensor data from at least one onboard vehicle sensor, the onboard vehicle sensor is a part of vehicle electronics of the host vehicle; obtaining edge sensor data from at least one edge sensor, the edge sensor is a part of an edge layer; generating a unified perception output based on the onboard vehicle sensor data and the edge sensor data; determining an AV action for the host vehicle based on the unified perception output; and providing the AV action to the host vehicle, wherein the host vehicle is configured to carry out the AV action.