LIDAR Localization Using Map Registration and Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle navigation systems face challenges in accurately determining a vehicle's location with respect to a map, especially when using uncalibrated LIDAR sensors, which can lead to insufficient accuracy for safe and comfortable autonomous operation.
Innovation Solution
A method that combines GPS, INS, and odometry data with uncalibrated LIDAR data by projecting reflectivity measurements onto a 2D grid, calculating gradients, and using normalized mutual information to register LIDAR data with map data, updated by an extended Kalman filter for improved location estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS, INS, and odometry data are used alone for location determination, then the system is simple to operate, but the location accuracy deteriorates to 8.5-11 meters
Solution Approach 1:
The patent combines multiple data sources (GPS, INS, odometry, and LIDAR) into a unified location determination system. The computing device integrates these diverse sensors and processes their data together through a common framework, achieving superior accuracy (within 20 cm) that none of the individual sensors could provide alone.
Solution Approach 2:
The patent introduces map data as an intermediary reference framework. The LIDAR data is registered against the map data, and this registered information serves as a mediator that connects the vehicle's sensor data to its actual location, enabling accurate position determination beyond what the sensors could achieve independently.
2Measurement precision
If uncalibrated LIDAR sensors are used, then the device complexity is reduced, but the location accuracy deteriorates and becomes insufficient for safe autonomous operation
Solution Approach 1:
The system performs self-calibration by using the map data as a reference truth. The computing device automatically registers the uncalibrated LIDAR data against the known map geometry, allowing the system to determine accurate location information without requiring manual calibration of the LIDAR sensors beforehand.
Solution Approach 2:
The patent transforms the LIDAR data from an uncalibrated state to a calibrated state through the registration process. By changing the reference frame and aligning the LIDAR measurements with the map data, the system effectively calibrates the sensor parameters dynamically during operation rather than requiring pre-calibration.
3Measurement precision
If multiple sensors and processing steps are combined to improve location accuracy, then the measurement precision improves to within 20 cm, but the device complexity increases
Solution Approach 1:
The patent segments the location determination process into distinct functional modules: GPS/INS/odometry processing, LIDAR data acquisition, map data retrieval, data registration, and final location calculation. This modular segmentation allows each component to be optimized independently while working together to achieve the overall accuracy goal.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances the accuracy of vehicle location determination from 8.5 to 11 meters to within 20 cm, enabling safer and more reliable autonomous vehicle operation.
Implementation Method 1
determining uncalibrated LIDAR data may include determining reflectivity measurements corresponding to ground-plane return data
Implementation Method 2
updated by an extended Kalman filter for improved location estimation
Data Source
AI summary
A system including a processor and a memory, the memory including instructions to be executed by the processor to determine map data, determine uncalibrated LIDAR data, determine a location of a vehicle in the map data by combining the map data with the uncalibrated LIDAR data, and operate the vehicle based on the location of the vehicle in the map data.