Autonomous Integration of Bench-Top Wet Lab Equipment

TL;DR

The paper tackles the challenge of affordable, flexible automation for bench-top lab equipment by targeting centrifugation with a computer vision pipeline and a low-cost gantry robot. It employs classical CV techniques—color thresholding for occupancy and circular Hough Transform for swing-bucket localization—coupled with a 3D-printed gantry and a Duet 2 controller to insert and remove tubes. Experimental results show ~5% detection error, 60% removal success, and 75% insertion success, with run-time dominated by serial communications but a fully autonomous workflow. This work demonstrates a practical pathway to integrating existing laboratory hardware in small labs, reducing manual labor and exposure to hazardous environments.

Abstract

Laboratory automation is an expensive and complicated endeavor with limited inflexible options for small-scale labs. We develop a prototype system for tending to a bench-top centrifuge using computer vision methods for color detection and circular Hough Transforms to detect and localize centrifuge buckets. Initial results show that the prototype is capable of automating the usage of regular bench-top lab equipment.

Figures (5)

• Figure 1: To automate a bench-top centrifuge, we used a 3D printer style gantry due to its simple control nature. We equipped the gantry system with a one degree of freedom end-effector and a Logitech C270 Webcam as shown in a). We chose to use the MiniPCR Bio Gyro Centrifuge as shown in b) because of its low-cost and wide availability. We equipped the centrifuge with a custom 3D printed swing bucket rotor to meet the required relative centrifugal force used in the prototypes DNA extraction protocol. To control the motors and sensors within the gantry system, we used the Duet 2 Ethernet micro-controller board Duet2, shown in c), because of its vast use in the 3D printer space and its expansive documentation.
• Figure 2: Diagram of the operations and decisions performed by the computer vision system.
• Figure 3: These images represent a sample output from the computer vision analysis system where the buckets are occupied by a yellow and orange test tube. a) This image shows the final bounding rectangles found after computing the contours found from the color detection system. b) This image shows the circular features and their center points found using the Hough Circle Transform, alongside the center points of the bounding rectangles found by the color detection.
• Figure 4: The robot system successfully removed the test tube sixty percent of the time. Fifteen percent of the trials resulted in an error due to improper localization, where the gripper was not properly lined up with the test tube. A total of ten percent of trials resulted in an error due to a failure by the computer vision system. Half of the computer vision errors were caused by the Hough Transform detecting a non-existent swing bucket and the other half were caused by improper test tube occupancy identification. The last fifteen percent of trials failed due to hardware-related interference.
• Figure 5: The robot system successfully inserted the test tubes seventy-five percent of the time. Five percent of the trials resulted in an error due to improper localization. Twenty percent of trials resulted in an error due to improper test tube occupancy identification.