Vehicle Pose Error Estimation Using Static Feature Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current localization methods for autonomous driving, such as GPS and inertial measurement units, fail to provide sufficient accuracy in complex urban environments, with errors exceeding 30 meters, which is inadequate for safe route planning due to the narrow lane widths of 3 to 4 meters.
Innovation Solution
A method involving an image capturing device on the vehicle to capture images, generate a reference image based on observed static features, and use a pose error network to implicitly compare the captured images with the reference images to determine corrections to the observed position and pose, thereby enhancing localization accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS and IMU sensors are used for localization, then real-time position determination is achieved, but localization accuracy deteriorates to 10 meters or worse in urban environments
Solution Approach 1:
The patent introduces static features (buildings, trees, signs) as intermediary objects between the vehicle and the localization system. These features serve as stable reference points that mediate the comparison between observed and reference images, enabling accurate position determination without relying directly on GPS/IMU signals that suffer from urban canyon effects
Solution Approach 2:
The patent replaces the mechanical/sensor-based localization system (GPS receivers, IMU accelerometers) with an optical comparison system. Instead of relying on satellite signals and inertial measurements, the system uses image capture devices to photograph static features and compares them with reference images from a database, substituting physical sensor measurements with visual pattern recognition
2Measurement precision
If image capture devices and pose error networks are used, then localization accuracy improves to 10 centimeters, but device complexity increases
Solution Approach 1:
The patent performs preliminary action by pre-capturing images of static features from known positions and storing them in a reference database before the actual localization task. This pre-processing creates a ready-to-use reference set that simplifies the real-time comparison process, allowing the system to achieve high accuracy without complex real-time computations
Solution Approach 2:
The patent creates copies of static feature images from known reference positions and stores them in a database. Instead of processing raw sensor data or performing complex geometric calculations, the system captures images, stores them as reference copies, and compares new images against these copies to determine position, significantly simplifying the computational complexity
Data Source
AI summary
The position and/or pose of a vehicle is determined in real time. An observed position and an observed pose of a vehicle are determined. A reference image is generated based on the observed position and the observed pose. The reference image comprises one or more reference static features. A captured image and the reference image are implicitly compared. Based on a result of the comparison, a correction to the observed position, the observed pose, or both is determined.