A digital twin framework for robust control of robotic-biological systems

Medical device regulatory standards are increasingly incorporating computational modelling and simulation to accommodate advanced manufacturing and device personalization. We present a method for robust testing of engineered soft tissue products involving a digital twin paradigm in combination with robotic systems. We developed and validated a digital twin framework for calibrating and controlling robotic-biological systems. A forward dynamics model of the robotic manipulator was developed, calibrated, and validated. After calibration, the accuracy of the digital twin in reproducing the experimental data improved in the time domain for all fourteen tested configurations and improved in frequency domain for nine configurations. We then demonstrated displacement control of a spring in lieu of a soft tissue element in a biological specimen. The simulated exper-iment matched the physical experiment with 0.09 mm (0.001%) root-mean-square error for a 2.9 mm (5.1%) length change. Finally, we demonstrated kinematic control of a digital twin of the knee through 70-degree passive flexion kinematics. The root-mean-square error was 2.00 degrees, 0.57 degrees, and 1.75 degrees degrees for flexion, adduc-tion, and internal rotations, respectively. The system well controlled novel mechanical elements and generated accurate kinematics in silico for a complex knee model. This calibration method could be applied to other sit-uations where the specimen is poorly represented in the model environment (e.g., human or animal tissues), and the control system could be extended to track internal parameters such as tissue strain (e.g., control knee liga-ment strain). Further development of this framework can facilitate medical device testing and innovative biomechanics research.

Automated summary

Medical device regulatory standards are incorporating computational modelling and simulation. A method for testing soft tissue products using a digital twin paradigm and robotic systems is presented. The accuracy of the digital twin in reproducing experimental data improved after calibration.