Dual-Robotic-Manipulator Collaborative System Based on Depth Image

Collaboration between robots, especially for multiple robotic manipulators, has been widely implemented in industrial applications attributed to its high efficiency and robustness working in harsh environments. However, successful applications of multiple agents in real-life laboratory applications are still quite limited to fulfilling the requirements of high precision and good adaptability when dealing with complex processes. This paper proposes a visual servo-controlled dual-arm robotic system handling routine experiments, such as gripping and placing tubes, screwing or unscrewing caps, shaking tubes, and pipetting. Based on the depth and RGB image feedback, the independently controlled UR3 robotic arms corporate efficiently and demonstrate the excellent potential to replace labor in human-defined specific tasks partially. We developed a program to detect the target and generate corresponding actions. As one of the typical applications, the laboratory-oriented configuration presented in this paper could adapt to other scenarios with minimal effort by customizing the end-effector according to different operation targets. Such a robotic-based scheme increases working efficiency and output value and demonstrates the capability of liberating the labor force from repetitive multi-process tasks. Systematic testing results have revealed that the success rate for tube grasp can reach 82.9%, and the success rate for a sequence of seven continuous movements is around 73.7%.