Adaptive robotic arm sensing for junction line extraction in footwear assembly

Abstract

This invention proposes an innovative method for identifying mold lines in footwear assembly. It requires only a one-time modeling of the shoe last needed for the manufacturing process, and can be applied to the identification of mold lines for diverse footwear styles. The method involves mounting a contour sensor on a robotic arm to capture the contour information of the assembled upper and sole, comparing it with the point cloud of the shoe last to identify the intersection of the sole and upper. Compared to existing methods, this invention significantly reduces modeling costs and effectively addresses the challenges posed by diverse footwear styles, production tolerances for sole and upper components, and process uncertainties, significantly improving the accuracy and efficiency of trajectory generation.

Claims

1. An adaptive robotic arm sensing method for identifying a mold line in footwear assembly, comprising: providing a robotic arm with a contour sensor disposed at the end of the robotic arm for identifying a mold line of a shoe having a shoe last; the mold line being a pressing line between an upper and a sole of the shoe; and correcting the relative position and orientation between the contour sensor and the robotic arm; The contour sensor is used to scan the shoe last to create point cloud data, and noise removal and normal vector calculation are performed. The contour sensor is moved to a position that can capture the contour of an assembly including the upper, sole, and shoe last. The normal vector of a sensing plane and the coordinates of a plane point of the contour sensor are converted into a unified coordinate system for processing. The distance between each point in the point cloud data and the sensing plane is calculated, and points falling within a distance threshold are captured and projected onto the sensing plane to obtain the contour of the shoe last on the sensing plane. After capturing the contour of the assembly using the contour sensor and converting it to the unified coordinate system, nearest neighbor matching is performed, and a pairing relationship between the shoe last, the assembly, and the point cloud data is established within the distance threshold. The optimal transformation matrix is ​​then calculated to rigidly align the point cloud data of the shoe last and the point cloud data of the assembly, and the overlapping points are obtained. The mold point of the cross section is determined by the principle of minimizing the distance between the intersection point of the sensing plane section and an adjacent mold point. After determining the position and direction of the next scanning plane based on the information of two adjacent mold points, several mold points are obtained and reconstructed into the mold line; and it is determined whether the distance and direction between the latest mold point and the starting mold point meet the preset closure conditions. If they meet the conditions, the mold point acquisition process is terminated and the mold line closure is completed.

2. The method as described in claim 1, wherein the calibration system of the contour sensor and the robotic arm uses a sphere of known radius as a calibration reference. The contour sensor on the robotic arm obtains the position of the sphere's contour information, and then controls the robotic arm to move the contour sensor along the direction of the robotic arm's reference coordinate system. By analyzing the coordinate change of the center of the sphere relative to the contour sensor's coordinate system during the movement, the relationship between the orientation of the contour sensor's coordinate system and the robotic arm's flange coordinate system can be obtained.

3. The method as described in claim 2, wherein the contour sensor is mounted on the robotic arm, the sphere is placed within the working range of the robotic arm, the robotic arm is then moved to a position where the contour sensor can capture the contour of the sphere, and the coordinates of the robotic arm relative to the reference coordinate system of the robotic arm and the relationship between the center of the sphere and the coordinate axis of the contour sensor are recorded. The contour sensor is used to capture the contour information of the sphere to obtain the coordinates of the center of the sphere and the radius of the cross-sectional circle. The distance between the center of the sphere and a contour sensing plane of the contour sensor is calculated. The robotic arm is controlled as the starting point, and the contour sensor is moved along the respective directions.

4. The method as described in claim 3, wherein after the contour sensor and the robotic arm are calibrated, the position of one of the die opening points on the contour sensing plane can be identified by the contour sensor on the robotic arm and converted into a workpiece coordinate system representation.

5. The method as described in claim 1, wherein the distance threshold for the point cloud data comparison is the fabric thickness of the upper adjacent to the sole.

6. The method as described in claim 5, wherein noise removal of the point cloud data includes noise reduction processing, calculating the average distance of points in the neighborhood using statistical filtering, and removing points that are greater than a threshold.

7. The method as described in claim 6, wherein the point cloud data is matched using the most recent iteration method to align the shoe last point cloud with the filtered assembly point cloud to determine the mold point position of the contour sensor sensing plane.

8. The method as described in claim 1, wherein since the outline of the shoe upper is not an assembly of the shoe last plus the shoe upper, the nearest iteration method is used to estimate the optimal coordinate transformation matrix so that the point cloud of the shoe last is aligned with the point cloud sensing plane of the shoe.

9. The method as described in claim 1, wherein the robotic arm is guided to two adjacent sensing planes, the sensing plane of the contour sensor being guided must contain only one die point, and after obtaining the die points of the two profiles, the next scanning plane is determined by calculating the vector of the two die points.

10. The method as described in claim 1, wherein the determination of the die point is based on the position of the intersection point in the cross section and the distance to the adjacent die point; and where there are multiple intersection points on a single sensing plane, a continuous die point trajectory is selected by the distance to the adjacent die point.

Images