Vision System for Detecting and Locating Micro-Scale Objects with Guided Cartesian Robot

This paper aims to use a vision system to detect and locate the position of microscale workpiece on the GelPak & wafer ring and command the end-effector to move to the targeted positions by using 2 sets of cameras. The object that is investigated in this design is Chip On-Sub-mount Assembly (COSA) with a footprint of ${270}\times {270}$ micrometers. The first camera is used to find the COSA's position and count the quantity of the COSA on the GelPak and wafer ring. The second camera is used to find the exact position of the COSA and then the end-effector is moved to the target position on COSA's sub-mount. The challenges in this process are accuracy, system integration, and the compliance of the GelPak and the wafer ring. With this proposed design, the performances of the system are examined and validated for the working prototype. The result of the wide camera for finding the approximate center position of the COSAs is 94% successful, the percentage of counting COSA is 90% accuracy. The XY position error from the given position compared to the center position of COSA's sub-mount is ${4}.{882}\ {\mu} {m}$ and ${8}.{206}\ {\mu}{m}$ for x and y axis respectively.