Dynamic Electromagnetic Navigation

Added: 2026-06-03

[00:00:00][Music] electromagnetic navigation systems represent a major breakthrough in medical robotics offering promising applications in targeted drug delivery and in guiding catheters by precisely adjusting magnetic fields to exert forces these systems enable accurate control of position and orientation of magnetic objects within the human body in this video we'll take a step back from medical applications and look at a more fundamental control problem the so-called inverted pendulum problem this setup allows us to demonstrate the dynamic capabilities of electromagnetic navigation systems making it an intriguing test bed for the study of Novel magnetic control algorithms to understand the complexity and challenge of controlling an inverted pendulum consider the analogy of trying to balance a long pole on the tip of your finger you can notice that the shorter the pendulum the more challenging it becomes to balance the setup consists of the Octo mag electromagnetic navigation system a magnetically driven arm the pendulum and our motion capture system which you can think of an indoor GPS so essentially the goal is to move the lower arm using the external magnetic field such that the upper pole remains balanced let's have a look you can see that the magnetically driven arm is making fine adjustments to keep the upper pole balanced you can also apply small nudges to the pendulum showing that the control algorithm can respond to unexpected disturbances of course the inverted pendulum is not clinically relevant in itself however such dynamically responsive algorithms could be beneficial for medical applications that demand precise real-time control potentially improving surgical outcomes to show you that we're not cheating let's switch off the feedback controller and let the pendulum drop and now let's also switch off the external magnetic field as we've shown earlier in this video the shorter the pendulum the harder it is to keep it balanced so let's see how robustly we can balance a 30 cm pole let's further reduce the length of the pole to 20 cm you can notice that it becomes very sensitive to disturbances next let's take a look at our offset compensation methods as you can see there is a deflection in the magnetically driven arm which is primarily caused by minor calibration errors in both the magnetic field and our motion capture system by leveraging online data we can actively learn the correct control actions to eliminate the steady state offset we can also follow trajectories while simultaneously balancing the poll to do this we utilize an iterative learning control scheme just as a human improves through repetition the iterative learning controller improves tracking performance from iteration to iteration to achieve that the learning controller veres online data and our Dynamic model to calculate a correction signal for the magnetic field for more information please follow the link to the publication in the description