The Recognition of Target and Attitude of Bucket for Excavator Robot Based on Color Mark Tracking

Abstract

The excavator robot need to track bucket goal and its attitude real-time and dynamically during it’s autonomous mining. On the basis of analyzing basic action of mining operations, collected sequence image of mining process of bucket. To reduce the effect of light environment to image processing, use the target tracking method based on color characteristics to recognize color mark block. The bucket attitude angle can be recognized according to the mark of angle attitude of bucket; the bucket target can be recognized according to the mark block of bucket target. Due to sheltered phenomenon, use stripe color block to recognize bucket target while mining process. The color mark block resolve the target tracking problem for bucket, but also fussing the information of pressure sensor of bucket, thus establishing the foundation for subsequent realization based on behavior of autonomous mining.

Calibration of visual measurement system for excavator manipulator pose

“…Visual measurement methods, in particular, have attracted attention from researchers [8]. Before 2018, studies used distinct selfmade markers with specific shapes and colors [9,10], or generic AprilTag markers [8,11,12], attached to buckets or manipulators for feature recognition, indirectly determining the manipulator's pose. Furthermore, some researchers have explored methods based on the contours of excavators [13][14][15][16].…”

“…These methods yield feature point pixel coordinates (FPPCs), and the process of generating the pose often involves mapping the relationship between FPPC and the corresponding manipulator pose, referred to as the 'pose mapping relationship'. Azar et al [11], Ma et al [27], and Wang et al [9] proposed a setup where the camera is fixed on the ground as shown in figure 1, the marker is attached to the excavator's manipulators, and the camera's optical axis is perpendicular to the working plane of the manipulators. In this scenario, FPPC can directly map the manipulator pose, and the pose mapping relationship is simple and requires no parameter calibration.…”