Sensor System Selection Using Error Curves for HD Map Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Map service providers face challenges in automatically selecting the most appropriate sensor system for high-definition map feature accuracy and reliability, as the positional quality of semantic features derived from sensor data is heavily dependent on the sensor system and number of observations, which is crucial for modern applications like autonomous driving.
Innovation Solution
A system that selects the most appropriate sensor system by calculating error data from multiple passes over known survey points, generating an error curve, and determining a target number of passes to meet accuracy and reliability specifications, using machine learning to compare sensor systems and provide output for selecting the best system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple sensor systems are used to collect data for high-definition maps, then measurement precision and reliability of map features improve, but device complexity and data processing requirements increase
Solution Approach 1:
The patent changes the parameter of sensor system evaluation from qualitative assessment to quantitative measurement by calculating error curves based on multiple passes over survey points. This allows objective comparison of different sensor systems using measurable parameters like error magnitude and convergence rate, resolving the complexity of selecting among multiple sensor systems.
Solution Approach 2:
The patent implements feedback by using calculated error data from multiple passes to generate error curves, which then inform the selection of appropriate sensor systems. The error specifications serve as feedback criteria to determine whether a sensor system meets the required accuracy thresholds for high-definition mapping applications.
2Reliability
If the number of passes over survey points is increased, then reliability and accuracy of sensor system evaluation improve, but loss of time and processing duration increase
Solution Approach 1:
The patent applies preliminary action by establishing error specifications and acceptance criteria before conducting the passes. This allows the evaluation process to be terminated early if a sensor system clearly meets or fails to meet the predetermined error specifications, reducing the number of passes needed while maintaining reliability.
Solution Approach 2:
The patent replaces the mechanical approach of increasing passes indefinitely with a mathematical model that predicts error convergence. By fitting error curves to the data and extrapolating to expected asymptotic behavior, the system can determine sufficient passes without requiring excessive actual passes, thus substituting computational analysis for extended physical testing.
3Manufacturing precision
If error specifications are made more stringent, then manufacturing precision of map features improves, but the target number of passes increases significantly
Solution Approach 1:
The patent applies dynamics by making the error specifications adaptive rather than fixed. The error thresholds can be adjusted based on the specific application requirements, sensor system capabilities, and survey point characteristics. This allows optimization between precision and productivity by setting appropriate error bounds for different mapping scenarios.
Solution Approach 2:
The patent applies local quality by allowing different error specifications for different types of survey points or geographic areas. Rather than applying a uniform error threshold across all locations, the system can impose stricter requirements on critical features or areas where high precision is essential, while accepting relaxed specifications in less critical areas, thus balancing overall precision with data collection efficiency.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An approach is provided for automatically selecting the most appropriate sensor system for high-definition map feature accuracy and reliability specifications. The approach, for example, involves selecting at least one survey point that has a known physical location. The approach also involves initiating a plurality of passes to capture a plurality of images of the at least one survey point using a sensor system. For each pass, the approach further involves calculating an estimated location of the at least one survey point based on the plurality of images and calculating error data based on the estimated location and the known location. The approach also involves generating an error curve with respect to a number of the plurality of passes based on the error data for said each pass. The approach further involves providing an output indicating a target number of passes to meet an error specification based on the error curve.