Robust Control of Heterogeneous Biohybrid Microrobots With Actuator Uncertainty Compensation

Biohybrid microrobots, which merge living biological matter with synthetic components, hold immense promise to revolutionize medicine by acting as targeted therapeutic agents. A fundamental barrier to their clinical translation, however, is their inherent biological heterogeneity - variations in size, morphology, and function - compounded by imperfections in actuation systems, resulting in unpredictable and unreliable motion. Here, we address this critical challenge by building a control framework that simultaneously compensates for both intrinsic robot-to-robot variability and extrinsic actuator nonlinearities. Our approach integrates real-time time-delay estimation to learn and cancel the unique, unmodeled dynamics of each individual microrobot, with a finite-time terminal sliding mode controller that ensures robust, high-fidelity trajectory tracking despite system imperfections. We demonstrate that this strategy standardizes the behavior of a heterogeneous population of cell-based microrobots (200-500 μm), reducing trajectory tracking errors by 51.3% compared to conventional controllers. By transforming these living constructs into reliable robotic agents, we enabled their precise deployment in a functional task, enhancing the closure rate of an in vitro tissue wound model by 77.8%. This work overcomes a crucial obstacle in biohybrid robotics, establishing a clear pathway toward harnessing the therapeutic potential of engineered living systems for applications in targeted drug delivery and regenerative medicine.