Actor-Based Map Attribute Fusion for Occluded Road Semantics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face inaccuracies and limitations in map data due to adverse weather, sensor blockages, and changes in road networks, necessitating improved methods for gathering and updating map attributes.
Innovation Solution
A computer-implemented method and system that utilize sensor data, perception data, and onboard map data to determine the vehicle's position, evaluate data via an online mapping module, and fuse detected map attributes with existing attributes, incorporating an actor-based reasoner module to infer semantic and geometrical road attributes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If map data is gathered using traditional sensor systems, then basic navigation functionality is achieved, but map data accuracy deteriorates due to adverse weather, sensor blockages, and road network changes
Solution Approach 1:
The patent introduces map actors (virtual representations of road elements like lane markings, curbs, and road boundaries) as intermediaries between the vehicle's sensor system and the map data. These actors serve as mediators that bridge the gap between limited sensor observations and comprehensive map understanding, allowing the system to infer map attributes even when direct sensor observation is blocked or unavailable due to adverse weather conditions
Solution Approach 2:
The system implements continuous feedback loops where map actors are updated based on sensor observations, and these updated actors subsequently improve map data accuracy. The perception system continuously observes road features, updates the corresponding map actors, and this feedback mechanism allows the system to adapt to road network changes and maintain accurate map data over time despite varying environmental conditions
2Productivity
If sensor systems operate in adverse weather conditions, then vehicle operation continues, but map data quality deteriorates due to blurriness and sensor blockages
Solution Approach 1:
The system pre-establishes map actors with expected road features and attributes before adverse weather conditions occur. These pre-defined actors serve as a buffer or cushion that maintains map data integrity even when sensor observations become blurry or blocked. The pre-existing structural knowledge in the map actors compensates for the degraded sensor input during adverse weather
Solution Approach 2:
Map actors serve as intermediary representations that decouple the vehicle's operational continuity from direct sensor quality. Even when sensors are blocked or produce blurry images during adverse weather, the map actors maintain a coherent representation of the road environment, allowing the vehicle to continue operating with reliable map data despite poor sensor conditions
3Duration of action of stationary object
If traditional map updating methods are used, then map data is maintained, but map accuracy deteriorates due to inability to detect changes in road networks
Solution Approach 1:
The system employs continuous feedback mechanisms where the perception system constantly monitors road features and compares them against the stored map actors. When discrepancies are detected—indicating road network changes—the map actors are updated accordingly. This feedback loop enables the system to maintain map data over time while simultaneously detecting and adapting to road network changes
Solution Approach 2:
The map updating process operates continuously rather than periodically, with the perception system constantly observing road features and updating map actors in real-time. This continuous action ensures that map data remains current with road network changes while maintaining stable map representation, resolving the contradiction between long-term map maintenance and change detection sensitivity
Data Source
AI summary
A system and method of actor-based map attribute generation and, more particularly, geometrical attribute and semantic attribute generation. A computer-implemented method that, when executed by data processing hardware, causes the data processing hardware to perform operations comprising gathering sensor data via a sensor system of a host vehicle, processing perception data via a perception system of the host vehicle, maintaining onboard map data of the host vehicle, determining a position of the host vehicle, evaluating the sensor data, the perception data, and the onboard map data via an online mapping module, and fusing one or more detected map attributes with existing map attributes.