Development of Automatic Reconfigurable Robotic Arms Using Vision-based Control
Zhang, M., Lotfi, N., Zhang, Z. & Esche, S. K.
Proceedings of the 2017 ASEE Annual Conference and Exposition, Columbus, Ohio, USA, June 25 - 28, 2017.
Abstract
The traditional industrial robotic systems are designed for mass production, in which each robot needs to be calibrated and programmed for a specific task.
These systems are expensive, only effective in specific applications, and vulnerable to any changes in the working environment or the task.
However, mass customization has become the new frontier in product manufacturing and marketing.
In order to satisfy the changes in market needs, especially for small and medium enterprises (SMEs), it is desirable to have low-cost industrial robotic systems that can be automatically reconfigured for different applications.
As a supplement to traditional robotic courses, students should be educated about how to design a robust, flexible, reconfigurable and redeployable industrial robotic system.
However, these contents are missing from most of current engineering curriculum due to the lack of appropriate educational robotic platforms.
The research presented here uses an assembly line robotic arm as a prototype to prove the feasibility of automatic reconfiguration.
The system first uses cameras to detect and recognize objects in the assembly line and then automatically chooses the best manipulator for the assembly task.
Next, the system predicts the end-effector’s error using cameras in a markerless approach.
The error is compensated in the last step, in which the system automatically generates the control commands for the robotic arm using visual results as feedback.
Using this robotic system as an educational platform, the students will be able to learn about several important aspects of flexible/reconfigurable manufacturing systems (e.g. robustness, flexibility, reconfigurability, redeployability, etc.) through one low-cost and easy-to-use experimental setup.
The research presented here uses an assembly line robotic arm as a prototype to prove the feasibility of automatic reconfiguration.
The system first uses cameras to detect and recognize objects in the assembly line and then automatically chooses the best manipulator for the assembly task.
Next, the system predicts the end-effector’s error using cameras in a markerless approach.
The error is compensated in the last step, in which the system automatically generates the control commands for the robotic arm using visual results as feedback.
Using this robotic system as an educational platform, the students will be able to learn about several important aspects of flexible/reconfigurable manufacturing systems (e.g. robustness, flexibility, reconfigurability, redeployability, etc.) through one low-cost and easy-to-use experimental setup.