Vehicle Trajectory Estimation Using Graph SLAM
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for creating and updating highly accurate digital maps for automated driving rely on dedicated mapping vehicles with high-cost and limited coverage, leading to potential delays in incorporating changes in the road network, especially when using sensor data with lower accuracy from conventional vehicles.
Innovation Solution
A device and method that estimates a vehicle's trajectory using sensor data from conventional vehicles, combining measured position values and odometry values to compensate for measurement errors, and employing a graph SLAM process to optimize the trajectory estimation, accounting for systematic and local errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dedicated mapping vehicles with high-accuracy sensors are used, then the accuracy of trajectory measurement is improved, but the cost and device complexity increase significantly
Solution Approach 1:
The patent introduces a backend server as an intermediary that receives low-accuracy sensor data from conventional vehicles, performs trajectory estimation using graph SLAM algorithms, and generates HD map updates. This mediator enables conventional vehicles to contribute to high-accuracy mapping without requiring them to possess complex high-precision sensors themselves.
Solution Approach 2:
The patent replaces the mechanical/sensor-based solution (dedicated mapping vehicles with lidar and high-accuracy GPS) with a computational approach using graph SLAM algorithms that process data from conventional sensors (camera, microphone, standard GPS, odometry) to achieve high-accuracy trajectory estimation.
2Measurement precision
If dedicated mapping vehicles are used, then the accuracy of HD map creation is improved, but the coverage area and frequency of map updates are limited
Solution Approach 1:
The patent enables conventional vehicles to perform mapping functions in addition to their primary transportation function. By processing sensor data from any conventional vehicle through the backend server's graph SLAM system, the system achieves universal coverage across the entire road network rather than being limited to routes taken by dedicated mapping vehicles.
Solution Approach 2:
Conventional vehicles automatically contribute their sensor data for trajectory estimation and HD map updates without requiring specialized equipment or dedicated operation. The vehicles serve themselves as mapping resources by generating data that is processed by the backend system, increasing both coverage and update frequency.
3Ease of manufacture
If sensor data from conventional vehicles is used, then the cost and accessibility are improved, but the measurement precision and resolution are reduced
Solution Approach 1:
The patent combines multiple low-precision sensor data sources (camera images, microphone audio, standard GPS positions, odometry from wheel sensors) into a unified trajectory estimation through graph SLAM processing. This fusion of diverse data types compensates for the individual limitations of each sensor, achieving high accuracy despite using only conventional, low-cost components.
Solution Approach 2:
The patent transforms the problem from one of improving individual sensor precision to one of optimizing the computational processing of multiple parameters. By changing the approach from hardware-based precision (high-accuracy sensors) to software-based precision (graph SLAM optimization across multiple data streams), the system achieves high accuracy using low-cost conventional sensors.
Data Source
AI summary
A device estimates an actual trajectory of a vehicle. The device is designed to determine a sequence of measured position values over a corresponding sequence of times by use of a position sensor. The device is further designed to determine a sequence of odometry values on the basis of sensor data from one or more vehicle sensors. The device is also designed to determine, as an estimation of the actual trajectory of the vehicle, a sequence of estimated pose values for the sequence of times and a systematic position offset for the sequence of measured position values so that an optimization criterion is improved, in particular is optimized at least locally.