ADAS Sensor Alignment Using Target-Based Pose Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine vision-based vehicle alignment systems are inefficient for large vehicles due to the need for rigid camera mounts, require time-consuming positioning procedures, and lack the ability to measure alignment without stopping, and they often require additional hardware for calibration, which increases costs and complexity.
Innovation Solution
A vehicle alignment system that uses a fixture with image sensors to capture data from targets on the vehicle and the floor, calculating the pose of these targets to generate instructions for positioning the fixture accurately without the need for rigid camera mounts or additional hardware, allowing for continuous measurement and alignment without stops.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If rigid camera mounts are used to maintain fixed relative positions, then measurement accuracy is improved, but device complexity and cost increase
Solution Approach 1:
The patent replaces the mechanical rigid mount system with a computational approach. Image sensors capture images of calibration targets, and a processor calculates the relative positions of cameras based on these images. This substitutes mechanical rigidity with optical measurement and mathematical computation, eliminating the need for physically rigid camera mounts while maintaining measurement accuracy.
Solution Approach 2:
The patent uses calibration targets with known geometries as reference copies. By capturing images of these standardized targets and comparing the captured geometry to the known geometry, the system determines camera positions and orientations. This copying approach allows accurate relative positioning without requiring physical rigid mounts between cameras.
2Measurement precision
If additional calibration hardware is added to determine camera positions, then measurement accuracy is improved, but device complexity and cost increase
Solution Approach 1:
The patent employs calibration targets with known geometries as reference copies. These targets are captured by the image sensors, and their known geometries are used to computationally determine camera positions and orientations. This approach replaces additional calibration hardware with standardized targets and mathematical computation.
Solution Approach 2:
The system uses the calibration targets and captured images to self-determine the relative positions of the cameras. The processor automatically calculates camera positions based on the captured target images and known target geometries, eliminating the need for external calibration hardware or manual setup procedures.
3Productivity
If continuous rotation measurement is implemented, then productivity is improved, but the need for no-stop positioning increases system requirements
Solution Approach 1:
The patent enables continuous rotation measurement by capturing images at multiple angles during wheel rotation without requiring stops. The system processes images captured throughout the rotation to calculate alignment parameters, maintaining continuous measurement capability. This eliminates the need for pause-and-measure procedures, improving productivity while using standard image capture technology.
Data Source
AI summary
Systems and methods are disclosed for calibrating and aligning automotive sensors, such as advanced driver assistance system (ADAS) sensors. Embodiments include a system having an image sensor mounted on a fixture in a known pose relative to the fixture. The image sensor is for viewing a target disposed on a vehicle and having a known pose relative to the vehicle, and for capturing image data of the target. A data processor is provided for performing the steps of calculating, using the image data, a pose of the target; calculating a pose of the image sensor relative to the vehicle using the calculated pose of the target; and generating instructions for positioning the fixture at a predetermined pose relative to the vehicle using the calculated pose of the image sensor relative to the vehicle.