Poster Presentation: Towards Tactile Intelligence for MIMMS
Date:
Presented a research poster titled “Towards Tactile Intelligence: Inverse Neural Modeling and Magnetic Field Sensing for MIMMS.”
Summary
This work explores how to enable tactile intelligence in medical robotics by combining:
- Inverse neural modeling
- Magnetic field sensing
- Soft material deformation analysis
Key Contributions
- Developed a reverse neural network (multi-head attention-based) to infer physical properties (Young’s modulus, length, cross-section, magnetic strength) from observed deformation.
- Designed a distributed dipole model with a ring-shaped magnetic sensor array (75 sensors) to reconstruct the 3D pose of flexible magnets.
- Achieved high-accuracy pose estimation with:
- < 3 mm position error
- < 5° orientation error
- Integrated AnySkin system + GeoSight camera for real-time visualization of deformation and force interaction.
Impact
This research contributes toward enabling robotic touch sensing in soft medical devices, improving grasping, perception, and interaction capabilities.
Poster PDF
Poster
Here is my picture for the poster presentation:
