Autonomous Vehicle Perception via Off-Vehicle Data Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous vehicle systems face limitations in range, field of view, and measuring errors when detecting objects, particularly pedestrians and cyclists, due to reliance on on-board sensors alone, which can lead to incomplete or inaccurate object classification and movement prediction.
Innovation Solution
The system integrates external data collection devices, such as mobile devices, to provide reported data on object location, type, and predicted trajectory, which is relayed to the autonomous vehicle via a server, allowing for correlation with on-board sensor data to enhance object detection and classification, and modify sensor models for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Length of stationary object
If on-board sensors alone are used for object detection, then device complexity is reduced, but detection range and field of view are limited
Solution Approach 1:
The patent combines on-board sensors with off-board collecting devices (mobile devices, roadside infrastructure) to create a distributed sensing network. This merging extends detection range and field of view while distributing system complexity across multiple devices rather than concentrating it all in the autonomous vehicle.
Solution Approach 2:
The patent introduces a server as an intermediary that receives data from collecting devices, processes it, and relays relevant information to the autonomous vehicle. This intermediary handles data management and processing complexity, allowing the vehicle to benefit from extended detection capabilities without bearing the full system complexity burden.
2Measurement precision
If on-board sensors alone are used for object detection, then device complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent merges data from multiple sources (on-board sensors and off-board collecting devices) to improve measurement precision. By combining multiple data streams and perspectives, the system achieves more accurate object classification and detection while distributing the processing complexity across the network.
3Area of stationary object
If external data collection devices are integrated, then detection range is extended, but device complexity increases
Solution Approach 1:
The patent segments the detection system into on-board sensors and separate off-board collecting devices. This segmentation allows the field of view to be extended across multiple spatial locations while distributing system complexity across independent devices, with the server coordinating them.
Solution Approach 2:
The patent employs universal collecting devices (mobile devices, roadside infrastructure) that can serve multiple functions: detecting objects, transmitting data, and potentially serving multiple autonomous vehicles. This multi-functionality extends field of view while avoiding the need for dedicated complex systems for each function.
4Measurement precision
If external data collection devices are integrated, then object classification accuracy is improved, but device complexity increases
Solution Approach 1:
The server acts as an intermediary that handles the complexity of data processing, correlation, and object classification. It receives raw data from multiple collecting devices, processes this information, and provides refined object classification results to the autonomous vehicle, improving accuracy while centralizing complexity management.
Data Source
AI summary
In an embodiment, a method includes receiving, at an autonomous vehicle, reported data regarding an object in proximity to the autonomous vehicle. The data is collected by a collecting device external to the autonomous vehicle, and is relayed to the autonomous vehicle via a server. The reported data includes a current location, type, or predicted location of the object. The method further includes determining whether the reported data of the object matches an object in an object list determined by on-board sensors of the autonomous vehicle. If the determination finds a found object in the object list, the method correlates the reported data of the object to the found object in the object list. Otherwise, the method adds the reported data of the object to an object list of objects detected by sensor from on-board sensors of the autonomous vehicle. In embodiments, the collecting device is a mobile device.