Remote Camera Pose Updates for GNSS-Denied Vehicle Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in achieving accurate localization in GNSS denial environments, such as indoor parking structures, where global navigation satellite systems cannot provide positioning data, hindering automated valet parking operations.
Innovation Solution
A system utilizing remote cameras to provide pose corrections, allowing the vehicle to determine an updated pose without expensive onboard sensors, enabling accurate vehicle localization in GNSS denial environments for autonomous operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GNSS receivers are used to achieve sub-meter localization accuracy, then localization precision is improved, but the system cannot operate in GNSS denial environments such as indoor parking structures
Solution Approach 1:
The patent introduces remote cameras as intermediary devices that capture images of the vehicle from external locations. These cameras act as mediators between the vehicle and the localization system, providing positioning data without requiring onboard sensors to be visible to satellites. The camera images serve as the intermediary mechanism that enables localization in GNSS-denied environments while maintaining accuracy.
Solution Approach 2:
The patent replaces the mechanical/optical system of GNSS satellite signal reception with an image-based system. Instead of relying on radio wave propagation from satellites, the system uses photographs captured by remote cameras to determine vehicle position. This substitution allows the system to function in environments where satellite signals are blocked by buildings or structures.
2Measurement precision
If expensive onboard sensors are deployed to achieve accurate localization in GNSS denial environments, then localization accuracy is improved, but system cost increases
Solution Approach 1:
The patent uses copies of the vehicle captured in photographs by remote cameras to determine position, rather than requiring expensive onboard sensors. The visual copy of the vehicle in the camera image serves as the basis for localization, eliminating the need for costly physical sensors on the vehicle itself. This copying approach provides accurate localization at lower system cost.
Solution Approach 2:
The patent replaces expensive, complex onboard sensor systems with a simpler, cheaper approach using remote camera images. The system uses inexpensive image processing and pose correction algorithms instead of costly specialized hardware, achieving comparable localization accuracy at reduced system cost.
3Measurement precision
If remote cameras with known poses are used to provide pose corrections, then localization accuracy is improved without expensive onboard sensors, but the system requires external infrastructure
Solution Approach 1:
The patent inverts the traditional localization architecture by placing the active sensing capability in fixed external cameras rather than in the moving vehicle. Instead of the vehicle carrying sensors that actively scan the environment, the environment contains passive cameras that capture images of the vehicle. This inversion simplifies the vehicle system while requiring infrastructure deployment.
Data Source
Figure 1
Figure 2-1
Figure 2-2
AI summary
This document describes techniques and systems for vehicle localization based on pose corrections from remote cameras in parking garages and other GNSS denial environments. A system can include a processor and computer-readable storage media comprising instructions that, when executed by the processor, cause the system to determine an estimated pose of the host vehicle within a GNSS denial environment after the host vehicle has been parked at a drop-off area. The system can also receive a corrected pose of the host vehicle from one or more remote cameras in the GNSS denial environment. The instructions further cause the processor to use the corrected pose to determine an updated pose for the host vehicle. In this way, the system can provide highly accurate vehicle localization in GNSS denial environments in a cost-effective manner to support automated valet parking and other autonomous driving functionalities.