The increasing investment in Minimal Invasive Surgery (MIS) has driven the development of medical devices that are progressively less invasive and more precise. This advancement has led to a new class of robots that are both highly dexterous and compliant with their operating environments, commonly referred to as Magnetic Actuated Soft Continuum Robot (MSCR). Magnetic actuation enables miniaturization while maintaining low stiffness, a combination that other types of Continuum Robot cannot achieve. However, these devices also introduce significant challenges. Their low stiffness, while beneficial for compliance, can hinder their progression through an endolumen unless complete shape control is maintained. Additionally, MSCRs are highly underactuated. This issue can be mitigated by pre-programming and analyzing the required shape configurations within the endolumen during the preoperative phase.
Previous formulations of MSCRs presented several limitations that reduced their practical effectiveness. Most approaches either proposed magnetization strategies that were not feasible for follow-the-leader motion or failed to utilize a continuously magnetized profile, restricting shape control. Additionally, prior mathematical models often relied on least-squares optimization, which does not guarantee that the computed states are physically achievable under actuation constraints, leading to inconsistencies between theoretical predictions and practical implementation.
This work addresses these shortcomings by employing a reduced-order continuum Cosserat rod formulation, enabling accurate state-dependent predictions of MSCR behavior. The proposed approach ensures optimal magnetic field actuation to support follow-the-leader motion using continuous magnetization profiles, enhancing precision in navigation. Furthermore, a metaheuristic optimization strategy is introduced, ensuring physically implementable, tailored solutions that significantly improve the adaptability of MSCRs in complex endoluminal pathways.
Experimental validations demonstrate the improved robustness, precision and adaptability of the proposed MSCR designs for navigating complex endolumens. However, environmental disturbances caused deviations from the predicted system evolution. To counteract these disturbances, a closed-loop feedback control strategy was implemented. Observing the robot shape in the workspace, the elasticity of the model was dynamically updated, allowing real-time adjustments to the actuation field to minimize shape errors. This approach effectively defines an adaptive closed-loop control system, enhancing stability and reducing drift. The feedback control strategy improved navigation performance by an average of 31% over open-loop configurations, with certain pathways demonstrating up to a 58% improvement.
By addressing existing limitations in shape control and environmental adaptability, this work advances MSCRs as viable solutions for challenging MIS scenarios, particularly in neurovascular, hepatobiliary, and pancreatobiliary interventions. It provides both a framework for the control and customization of these otherwise highly underactuated systems and introduces an adaptive feedback law to mitigate environmental disturbances, improving reliability in real-world applications.
