Core Concepts
CRESSim is a realistic surgical simulator based on Unity and PhysX 5 that enables the simulation of various contact-rich manipulation tasks in surgeries, including tissue grasping and deformation, blood suction, and tissue cutting.
Abstract
The authors present CRESSim, a new surgical simulation platform built on Unity and the PhysX 5 physics engine. CRESSim is designed to enable the simulation of diverse types of objects and contact-rich manipulation tasks typically present in surgical environments, such as soft tissue, fluids, and cloth.
Key highlights:
CRESSim supports the simulation of rigid bodies, soft bodies using FEM, cloth and fluids using PBD, and serial robots using articulation joints.
The platform incorporates the real-world da Vinci Research Kit (dVRK) console and master tool manipulator (MTM) robots, allowing for VR-based teleoperation of the simulated surgical scene.
Three example surgical tasks are demonstrated: tissue grasping and deformation, blood suction, and tissue cutting. These tasks showcase the simulator's ability to handle complex contact-rich manipulation involving various surgical instruments, soft tissue, and body fluids.
Preliminary experiments and profiling show the platform's capability to simulate surgical tasks and allow real-time teleoperation, with room for further optimization.
The authors aim to enhance the simulator to cover more realistic surgical scenes and instruments in future work, with the long-term goal of providing an open-source research platform for surgical robotics applications.
Stats
Tissue grasping and deformation scene:
Physics advance time: 9.51 ± 0.63 ms
Mesh and particle post-processing time: 0.12 ± 0.02 ms
Total FixedUpdate time: 9.68 ± 0.63 ms
Blood suction scene:
Physics advance time: 14.00 ± 0.31 ms
Mesh and particle post-processing time: 0.19 ± 0.03 ms
Total FixedUpdate time: 14.24 ± 0.31 ms
Tissue cutting scene:
Physics advance time: 15.86 ± 0.39 ms
Mesh and particle post-processing time: 0.39 ± 0.16 ms
Total FixedUpdate time: 16.52 ± 0.81 ms