In vivo direct drug targeting aims at delivering drug molecules loaded on microrobots to the diseased site using the shortest possible physiological routes, which potentially improves targeting efficiency and reduces systemic toxicity. It is thus essential to consider realistic in-body limitations for direct drug targeting applications. Here, we present a novel controller for microrobot maneuver by considering four key in vivo constraints: non-Euclidean structure of capillaries, irreversibility of blood flow, invisibility of microvasculature, and inaccuracy of microrobot tracking. We use the taxicab geometry of capillaries as the a priori knowledge for steering and tracking a microrobot in lattice-like vessels. Furthermore, we introduce a minimax repulsive boundary function to prevent the microrobot from getting too close to the boundaries imposed by the direction of blood flow. We also propose a novel Kalman filtering algorithm to reduce tracking error, while avoiding possible obstacles such as vessel walls without knowing their actual locations. The proposed control method consists of four modules, namely a model predictive control module for tumor targeting, a Kalman filtering module for microrobot tracking, a blind obstacle detection module, and a vessel structure estimation module. The interplay of these four modules offers successful maneuver and tracking of the microrobot while avoiding obstacles in a blind manner by utilizing the taxicab geometry of blood vessels. We present a comprehensive in silico simulation study to verify our designed controller.

中文翻译：

---

使用出租车几何学进行直接药物靶向的微型机器人体内控制和跟踪的可行性研究

体内直接药物靶向旨在使用最短的生理途径将装载在微型机器人上的药物分子输送到患病部位，这有可能提高靶向效率并降低全身毒性。因此，必须考虑直接药物靶向应用的实际体内限制。在这里，我们通过考虑四个关键的体内约束条件来提出一种用于微型机器人操纵的新型控制器：毛细血管的非欧式结构、血流的不可逆性、微血管系统的不可见性和微型机器人跟踪的不准确性。我们使用毛细血管的出租车几何形状作为先验知识，用于在格状容器中操纵和跟踪微型机器人。此外，我们引入了一个极小极大排斥边界函数，以防止微型机器人过于靠近血流方向所施加的边界。我们还提出了一种新颖的卡尔曼滤波算法，以减少跟踪误差，同时在不知道其实际位置的情况下避免可能的障碍物，例如血管壁。所提出的控制方法由四个模块组成，即用于肿瘤靶向的模型预测控制模块、用于微型机器人跟踪的卡尔曼滤波模块、盲障碍物检测模块和血管结构估计模块。这四个模块的相互作用提供了微型机器人的成功操纵和跟踪，同时通过利用血管的出租车几何形状以盲目的方式避开障碍物。我们提供了全面的计算机模拟研究来验证我们设计的控制器。