Simulation for surgical robotics
GPU-accelerated soft-body mechanics, differentiable physics, and graph-based models for surgical-robotics research at Johns Hopkins.
Research question
How can simulation connect models of soft tissue with observations of the physical world?
Motivation
Soft-body simulation gives surgical-robotics research a computational representation of deformable material. The work recorded here connects that representation with real-world displacement information.
Approach
The supplied research record includes GPU-accelerated XPBD simulation using NVIDIA Warp, real-time skin deformation, differentiable physics, and graph neural networks. Real-world displacement information is used for calibration.
What was built
The documented work includes a soft-body simulation system and real-time skin-deformation simulation. Further implementation details, diagrams, and reproducible experiments have not yet been added to this public note.
Results
No verified numerical results are recorded here yet.
Open questions
The broader areas of interest include autonomous surgical robotics, robot learning, and sim-to-real transfer. Specific experimental questions will be added alongside their supporting evidence.
References
Paper, code, authorship, and exact project dates are not yet supplied.
XPBD · NVIDIA Warp · Soft-body mechanics · Differentiable physics · GNNs · Real-world calibration