Sensor Calibration Using 3D Shape Data for Mining Vehicles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge lies in accurately calibrating sensors on large vehicles like mining dump trucks, especially when assembled outdoors on uneven terrain, where precise landmark installation is difficult due to environmental conditions and human skill limitations, affecting the reliability of obstacle detection systems.
Innovation Solution
A calibration system that utilizes three-dimensional shape information to determine the relative position and orientation of vehicle-mounted sensors, employing units such as three-dimensional shape information acquiring, vehicle sensor position detecting, and landmark associating units to accurately measure and correct sensor positions and orientations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a plurality of sensors are combined for obstacle detection, then obstacle detection performance is improved, but calibration accuracy between sensors deteriorates when measurement regions are not superimposed
Solution Approach 1:
The patent transitions from two-dimensional planar calibration methods to three-dimensional spatial calibration. By acquiring three-dimensional shape information of the vehicle body and using it to calculate sensor installation positions and landmark positions in 3D space, the system enables accurate calibration even when sensors are positioned in different measurement regions (front, rear, left, right) of the vehicle, resolving the limitation of traditional superimposed measurement region requirements.
2Ease of manufacture
If landmarks for calibration are installed manually on uneven terrain, then calibration can be performed, but calibration accuracy deteriorates due to ground leveling issues and human skill limitations
Solution Approach 1:
The patent replaces manual mechanical measurement and installation methods with automated three-dimensional shape information acquisition and processing. Instead of relying on manual landmark installation and measurement on uneven terrain, the system uses 3D scanning or imaging to automatically capture the vehicle body shape, calculate precise sensor and landmark positions, and perform calibration computations, thereby eliminating the need for manual intervention and ground leveling.
3Area of stationary object
If sensors are attached to large vehicles like mining dump trucks, then obstacle detection coverage is improved, but calibration difficulty increases due to vehicle size and aging-induced positional deviation
Solution Approach 1:
The patent changes the calibration approach from fixed factory calibration to dynamic recalibration based on three-dimensional shape information. By continuously acquiring 3D data of the vehicle body and recalculating sensor installation positions and landmark positions, the system compensates for aging-induced positional deviations and maintains calibration accuracy despite the large vehicle size and extended operational periods.
Data Source
AI summary
When a combination of a plurality of sensors is used for obstacle detection, the present invention is to detect the relative positions of the sensors, to correct inter-sensor parameters, and to provide accurate obstacle detection. This calibration system is provided with a first landmark detecting unit for detecting the position of a first landmark from three-dimensional shape information; a landmark associating unit for determining a correspondence relation, between the first landmark position detected by the first landmark detecting unit and an attachment position to the vehicle of the first landmark estimated by a vehicle landmark relative position estimating unit; and a vehicle sensor relative orientation estimating unit for estimating the relative position and orientation of the vehicle and a first measurement section based on the information from the vehicle landmark relative position estimating unit and a landmark associating unit.