Sparse Radar Localization for GPS-Denied Autonomous Vehicles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Localizing autonomous vehicles in GPS-denied environments and adverse conditions such as dust, fog, snow, or rain is complex due to the lack of infrastructure and unreliable sensor data.
Innovation Solution
The method involves receiving radar data, filtering out motion-related points, combining it with previous data using vehicle velocity, and matching it with map data using algorithms like Kalman or particle filters to estimate the vehicle's location, which can be done without GPS, using radar sensors that penetrate adverse conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If radar data is used for localization in GPS-denied environments, then reliability is improved, but measurement precision deteriorates due to sparse data points and motion-related errors
Solution Approach 1:
The patent extracts and removes motion-related radar data points from the dataset before performing localization matching. By identifying and eliminating points representing moving objects (using velocity thresholds or clustering), the system prevents these erroneous points from degrading the precision of position estimation, thereby resolving the contradiction between using radar for reliable GPS-denied localization and maintaining measurement precision despite sparse and noisy data
Solution Approach 2:
The patent performs preliminary actions by combining current radar data with previous radar data and velocity information before the actual localization matching. This pre-processing includes estimating vehicle pose using Kalman filters or particle filters, and preparing cleaned radar point clouds in advance, which improves measurement precision while maintaining the reliability benefit of using radar in GPS-denied environments
2Length of stationary object
If radar sensors are used instead of other sensors, then detection range is improved, but device complexity increases due to data processing requirements
Solution Approach 1:
The patent extracts only the necessary radar data points for localization by removing motion-related points and combining with previous frames. This selective extraction reduces the data processing burden while maintaining the long detection range advantage of radar sensors, effectively resolving the contradiction between extended detection range and processing complexity
Solution Approach 2:
The patent performs preliminary data processing by combining current and previous radar data, removing erroneous points, and estimating vehicle pose before final localization matching. This staged approach breaks down complex processing into manageable steps, reducing overall system complexity while preserving radar's long-range detection capability
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 enables accurate localization of autonomous vehicles in environments without infrastructure, providing reliable positioning and orientation even in harsh conditions, leveraging radar's ability to penetrate dust, fog, and rain, and offering a cost-effective solution with a longer detection range compared to other sensors.
Implementation Method 1
radar sensors, for example, can penetrate dust, fog, falling snow, and/or falling rain
Implementation Method 2
identifying radar data points in the current radar data that represent objects in motion
Implementation Method 3
estimating an estimated location of the autonomous vehicle based on previous radar data and vehicle velocity data comprises using a Kalman filter or any variant of a Kalman filter, or a particle filter or any variants of a particle filter
Data Source
AI summary
An autonomous vehicle is disclosed that includes a velocity sensor; a radar system; a digital storage medium; and a controller in communication with the digital storage medium, radar system, and the velocity sensor. The controller, for example, may retrieve map data from the digital storage medium; receive current radar data from the radar system; receive autonomous vehicle velocity data from the velocity sensor; identify radar data points in the current radar data that represent objects in motion; remove radar data points from the current radar data that represent objects in motion; and/or match the radar data with the map data to return location data.