Redundant Vehicle Localization Pipelines for Drift-Resilient Positioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current localization systems for self-driving vehicles, such as dead-reckoning, LiDAR, and radar-based systems, face challenges like drift, temporary 'blips,' and security concerns, leading to inaccuracies and single points of failure, which compromise safety and accuracy.
Innovation Solution
Implementing multiple redundant localization pipelines that combine sensor data from distinct sources, like LiDAR, radar, and dead reckoning, to validate and weight location predictions, ensuring accuracy and resilience by setting thresholds for agreement among pipelines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple redundant localization pipelines are implemented, then reliability is improved, but device complexity increases
Solution Approach 1:
The localization system is divided into multiple independent pipelines (LiDAR-based, radar-based, dead-reckoning-based) that each process sensor data separately. Each pipeline operates autonomously to determine vehicle location, and their outputs are subsequently combined through validation and weighting mechanisms. This segmentation allows the system to achieve redundancy and improved reliability while maintaining manageable complexity through modular architecture.
2Measurement precision
If multiple sensor types are combined, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The system merges outputs from multiple independent localization pipelines that use different sensor types (LiDAR, radar, dead reckoning). Each pipeline processes data from its specific sensor modality separately, and the final location determination combines these independent results through validation checks and weighted aggregation. This merging approach improves measurement precision by leveraging complementary sensor strengths while avoiding the complexity of directly fusing raw sensor data.
Data Source
AI summary
Provided are methods for semantic annotation of sensor data using unreliable map annotation inputs, which can include training a machine learning model to accept inputs including images representing sensor data for a geographic area and unreliable semantic annotations for the geographic area. The machine learning model can be trained against validated semantic annotations for the geographic area, such that subsequent to training, additional images representing sensor data and additional unreliable semantic annotations can be passed through the neural network to provide predicted semantic annotations for the additional images. Systems and computer program products are also provided.