Adaptive Vision Localization for Industrial Robots

Abstract

The traditional serial type articulated robot has good positioning repeatability; it is still difficult to have good positioning accuracy. Unfortunately, accuracy is a necessary condition for aviation applications. Under the development of aviation heavy industry, laser tracker is used to promote the accuracy of the robot. However, there are some limitations about using laser odometry in practice: 1. complicated pre-processing and calibration 2. Limited to plain and simple environment 3. Expensive manufacture and maintenance. Therefore, this study developed a visual and inertial system to achieve the objective without less cost. To this end, a light-weight and low-cost visual-inertial localization method is applied to achieve the requirement of the high accuracy of the industry robot. Final, the past two methods and results (Denavit–Hartenberg model calibration and iterative learning control) to improve robot location accuracy are also included in our comparison. The results show that the proposed method has sufficient performance to meet most applications.