Teleoperation in surgery is a bit like trying to tie your shoelaces with chopsticks while wearing thick winter gloves—you can see what you're doing, but you can't feel it. This "sensory gap" forces surgeons to rely entirely on visual cues, which can lead to increased tissue damage and longer procedure times. While high-end commercial platforms are beginning to reintroduce force feedback, they remain "black boxes" that are off-limits to researchers.
A new paper titled "An Open-Source ROS 2 Framework for Shared Control in Surgical Teleoperation" aims to break down these walls. The researchers present a modern, open-source bridge between the compact Haply Inverse3 haptic device and the da Vinci Research Kit (dVRK), providing a playground for the next generation of haptic shared-control algorithms.
TL;DR
Researchers have developed an open-source, ROS 2-native framework that integrates the affordable Haply Inverse3 haptic controller with the da Vinci Research Kit. Their study shows that even simple haptic "guidance" significantly reduces errors and task time, providing a reproducible platform for open surgical robotics research.
What Problem Does This Paper Address?
In robotic surgery, "shared control" is a paradigm where the robot doesn't just follow the surgeon's lead but actively assists—for example, by providing haptic "virtual fixtures" that gently nudge the surgeon back onto a safe path.
The challenge is that the "gold standard" haptic devices used in research are often bulky and expensive. Furthermore, most existing software integrations rely on legacy middleware. There is a pressing need for a modern (ROS 2), accessible, and open-source way to connect compact haptic hardware with surgical robots so that researchers can experiment with new ways to help surgeons "feel" their way through a procedure.
Methods
The team built a complete software stack (released under the MIT license) to facilitate this integration. Key components include:
- ROS 2 Driver: A native interface for the Haply Inverse3 and its VerseGrip stylus, which handles high-frequency telemetry over WebSockets.
- 6-DOF Mapping: Since the Inverse3 only tracks 3-DOF translation, the researchers combined it with the separately tracked VerseGrip orientation to create a full 6-DOF command pipeline for the da Vinci's patient-side manipulators (PSM).
- Haptic Guidance Modes: They implemented two types of "virtual fixtures" (see Figure 2):
- Mode 1 (Linear): A "spring-like" force that increases linearly as the surgeon deviates from the desired path.
- Mode 2 (Progressive): A piecewise-linear profile that stays "soft" near the path to preserve dexterity but ramps up stiffness quickly as the surgeon nears a boundary.
Datasets and Benchmarks
The framework was evaluated using a "hot-wire" benchmark task. Much like the classic "Operation" board game, users had to move a ring along a multi-segment wire without making contact.
- Simulation: A within-subjects study with 20 participants was conducted in a simulated RViz environment.
- Physical Hardware: The system was also validated on actual dVRK hardware to prove real-world feasibility.
Key Insights and Ideas
- Simplicity is Robust: Surprisingly, the simpler Linear Guidance (M1) outperformed the more complex Progressive Stiffness (M2). The authors suggest that for novice operators, the sudden change in force in M2 might increase cognitive load rather than help.
- The "Sensory Gap" is Real: Participants using haptic guidance reported feeling more in control and more satisfied with their performance, even when the visual feedback was limited to a 2D monitor.
- Openness as a Contribution: By using ROS 2 and releasing the code, the authors have lowered the barrier to entry for other labs to study human-robot interaction in surgery without needing proprietary (and expensive) hardware-software bundles.
Key Results
The user study yielded clear quantitative benefits for haptic assistance:
- Fewer Errors: Linear guidance (M1) reduced the number of wire contacts from an average of 31.85 (manual control) down to 20.90.
- Faster Completion: Task time dropped significantly from 96.31 seconds to 75.01 seconds with linear guidance.
- Technical Feasibility: The physical integration achieved a stable 1 kHz control loop, matching the dVRK's native rate and ensuring smooth teleoperation.
Limitations and Future Work
While promising, the authors note a few caveats:
- 2D vs. 3D: The study used a 2D monitor. In real surgery, surgeons use 3D binocular displays, which might reduce the relative importance of haptic cues.
- Novice Users: The participants were not trained surgeons. Expert surgeons might react differently to haptic "nudges."
- Task Complexity: The hot-wire task is a "primitive" teleoperation benchmark. Future work will involve more complex, surgically representative tasks like suturing or tissue dissection.
The team plans to integrate stereoscopic VR and explore adaptive shared control, where the robot adjusts its level of assistance based on the user's current cognitive load or skill level.
Check out the code: The framework is available on GitHub at ABC-iRobotics/haply_ros2_interface.
Summary updated