Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Controlling magnetic objects in space through learning-based sensing
The goal of the project is to control magnetic objects in space through, utilizing high-frequency magnetic fields for localization.
Keywords: control, machine learning, state estimation, electromagnetic induction
Magnetically actuated continuum robots possess several desirable properties that make them promising for a variety of medical
applications. They are characterized by low-weight, inherent compliance, and high flexibility which - for instance - allows them to be
deployed in minimally invasive surgeries. At the Multi-Scale Robotics Lab at ETH Zürich, we have developed several so-called electroMagnetic Navigation Systems (eMNS) throughout the past decade. Simply speaking, an eMNS is an array of electro-magnetic coils that are
arranged in a specific configuration. By applying an electric current to the coils, a magnetic field is generated that can be used to control
magnetic objects in the space. In addition, a superimposed localization field with a unique frequency can be generated throughout the
space by each electromagnet. The localization field of the electromagnets can be sensed by pickup coils integrated into the device to be
actuated which allows for estimation of the 3D position and shape of the continuum robot.
Magnetically actuated continuum robots possess several desirable properties that make them promising for a variety of medical applications. They are characterized by low-weight, inherent compliance, and high flexibility which - for instance - allows them to be deployed in minimally invasive surgeries. At the Multi-Scale Robotics Lab at ETH Zürich, we have developed several so-called electroMagnetic Navigation Systems (eMNS) throughout the past decade. Simply speaking, an eMNS is an array of electro-magnetic coils that are arranged in a specific configuration. By applying an electric current to the coils, a magnetic field is generated that can be used to control magnetic objects in the space. In addition, a superimposed localization field with a unique frequency can be generated throughout the space by each electromagnet. The localization field of the electromagnets can be sensed by pickup coils integrated into the device to be actuated which allows for estimation of the 3D position and shape of the continuum robot.
Tasks:
• Sensing & Data Acquisition: Create pickups coils and implement analog signal conditioning and sampling (electronic simulation with LTSpice,
electronic prototyping, Python programming)
• Magnetig field modelling: Test & evaluate different models for predicting the localization field in space (e.g. machine learning, tricubic
interpolation, multipole expansion etc.)
• State estimation: Given measurements from the pickup coil, estimate the state of the continuum robot using the magnetic field model.
Implementation of a Kalman Filter to improve tracking capabilities.
• Control: Use the state estimate to control the motion of the continuum robot in space
• Inverted Pendulum Control: Use the state estimate to balance an already working 3D inverted pendulum system, to demonstrate the dynamic
capabilities of the pick up coil bases state estimate, as shown in this video (https://youtu.be/fNWS-9-lD84)
Tasks: • Sensing & Data Acquisition: Create pickups coils and implement analog signal conditioning and sampling (electronic simulation with LTSpice, electronic prototyping, Python programming) • Magnetig field modelling: Test & evaluate different models for predicting the localization field in space (e.g. machine learning, tricubic interpolation, multipole expansion etc.) • State estimation: Given measurements from the pickup coil, estimate the state of the continuum robot using the magnetic field model. Implementation of a Kalman Filter to improve tracking capabilities. • Control: Use the state estimate to control the motion of the continuum robot in space • Inverted Pendulum Control: Use the state estimate to balance an already working 3D inverted pendulum system, to demonstrate the dynamic capabilities of the pick up coil bases state estimate, as shown in this video (https://youtu.be/fNWS-9-lD84)
Denis von Arx, Jasan Zughaibi
Please email your CV and transcripts to dvarx@ethz.ch and zjasan@ethz.ch
Denis von Arx, Jasan Zughaibi Please email your CV and transcripts to dvarx@ethz.ch and zjasan@ethz.ch