Computer Vision Guided Robotic Arm
Maximized process efficiency by 38%
Overview
An LED manufacturing company hired us to design and build a computer vision system for an existing robotic arm. The computer vision system was designed to pick and place specific PCB components into predetermined locations.
Solution
Our solution aligned the camera field of view with robotic arm kinematics to reliably pick and place components on to a series of PCBs. The primary challenge faced was system repeatability, which was solved using a machine learning algorithm that would learn the mechanical shift and factor it into movement.
Results
The system was fully scaled to 5 PCB assembly stations reducing employee turnover and maximized employee and process efficiency by 38%.
Tools Used
Dorna robot
Intel Realsense D435
OpenCV
Advanced Algorithm Development
Robotic Kinematics