3D Sensor Registration Using Robot Joint Displacement Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current registration algorithms for 3D data in factory automation require extensive calculations and are not robust, leading to lengthy processing times and potential registration failures, especially when initial shifts between point clouds are large.
Innovation Solution
A measurement system that synchronizes the 3D sensor's data capture with displacement detection of robot joints only at specific points, reducing the need for frequent robot motion stops and utilizing coordinate transformation and position/orientation estimation units to register 3D data efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If registration algorithms are used to minimize distances between corresponding points of multiple point clouds, then measurement precision is improved, but processing time increases significantly
Solution Approach 1:
The patent applies preliminary action by performing coarse registration using robot motion data before executing fine registration with ICP. The coarse registration pre-aligns point clouds based on robot joint displacements, reducing the initial distance between corresponding points. This preliminary alignment minimizes the computational burden and iteration requirements of the subsequent ICP fine registration, thereby reducing overall processing time while maintaining high registration precision.
2Measurement precision
If robot stops at every measurement point to synchronize timing, then measurement precision is improved, but productivity decreases
Solution Approach 1:
The patent extracts the synchronization requirement from every measurement point and applies it only to the specific measurement point where the robot stops. By separating the synchronization need from the general measurement process, the system achieves precise timing alignment where necessary (at the stopped measurement point) while maintaining continuous robot motion at other points, thus preserving productivity without sacrificing measurement precision.
Solution Approach 2:
The patent introduces robot joint displacement data as an intermediary to bridge the timing gap between continuous robot motion and discrete measurement points. Instead of stopping the robot at every point for synchronization, the system uses displacement data from robot joints as a mediator to calculate and compensate for positional differences, enabling accurate registration without frequent interruptions to robot motion.
3Reliability
If coarse registration is performed as preprocessing, then reliability of fine registration is improved, but device complexity increases
Solution Approach 1:
The patent segments the registration process into two distinct stages: coarse registration and fine registration. Coarse registration handles the initial alignment using robot motion data, while fine registration refines the alignment using ICP. This segmentation improves reliability by ensuring that fine registration starts from a pre-aligned state, reducing the risk of convergence to local minima. The complexity is managed by clearly defining the boundaries and data flow between the two segments.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
Provided are a measurement system, a measurement device, a measurement method, and a measurement program for reducing the time needed to perform registration of 3D data of a measurement object. 3D data (110-2) is registered to 3D data (110-1) based on the displacements of joints of a robot (60) at a point in time when a 3D sensor (70) measures 3D data (110-1) of a measurement object (80) at a specific measurement point (90-1) while the robot (60) is stopped, and the displacements of the joints of the robot (60) at a point in time when the 3D sensor (70) measures 3D data (110-2) of the measurement object (80) at a measurement point (90-2) other than the specific measurement point (90-1) while that robot (60) is in motion. The 3D data (110-2) is further registered to the 3D data (110-1) such that a registration error between the 3D data (110-1) and the 3D data (110-2) is less than a threshold value. Similarly, each of 3D data (110-3, 110-4, ..., and 110-N) is registered to the 3D data (110-1).