A data-driven approach to optimal control and motion planning of medical nanorobots with linear quadratic regulator control for targeted drug delivery

Abstract

As a minimally invasive medical technique, nanorobots provide practical solutions to medical applications where, for instance, traditional surgery or cancer treatment may pose severe risks due to the delicate nature of said medical procedure. In specific cases where the intended area of surgery is difficult to reach (e.g., gastrointestinal tract, blood-brain barrier), nanorobots can be guided toward such target sites for various purposes, including drug delivery and reduction/elimination of cancer cells. Current literature relating to medical nanorobotics explore propulsion techniques, nanorobot design, and clinical testing of nanorobots in animals that suffer from various diseases. However, spare research has been conducted on the control of nanorobots. Since robots for targeted drug delivery are micro- or nanometers in size and rely on unconventional propulsion, implementing control laws become vital to ensure the robot reaches its target. Therefore, this research explores optimal control and tracking of nanorobots. Though limited, existing literature regarding nanorobotic control explore nonlinear control methods. In various publications, the model-dependent Model Predictive Control (MPC) is implemented by deriving nonlinear control laws via backstepping approaches. With nonlinear controllers, excellent robust tracking and control have been demonstrated with low tracking errors. However, as will be discussed in the following chapters, the obtained results come with computational costs and the high complexity of derived control laws. For this reason, this research proposes a linear control approach that can provide a good balance between computation and control performance. Further, linear control laws are guaranteed, and thus, backstepping is eliminated, provided that dominant dynamical information is retained after linearization.