VISTA benchmark shows network lag slashes surgical robot success rates to 12%
GEO satellite links cause 88% failure in remote surgery tasks, study finds
A new benchmark called VISTA, presented at ICRA 2026, systematically quantifies how network impairments degrade surgical teleoperation performance. The researchers simulated five realistic network conditions using Linux Traffic Control with NetEm and a Gilbert-Elliott loss model: Hospital LAN, 5G Urban, 4G Rural, LEO Satellite, and GEO Satellite. Over 375 standardized peg transfer trials, they measured network QoS (packet loss, latency, jitter), objective video quality (PSNR, SSIM, VMAF), and temporal continuity (freeze rate), while maintaining a stable reverse control channel.
The results are stark: success rates plummet from 97% under Hospital LAN to just 12% under GEO Satellite connectivity. Even relatively modern 4G Rural networks saw only 35% task success, while 5G Urban managed 79%. Mean task completion times for successful attempts nearly tripled from 80 seconds (Hospital LAN) to 255 seconds (GEO Satellite). These findings underscore that video quality degradation directly impacts human task performance in remote surgery. VISTA's open-source code provides a reproducible foundation for testing video streaming stacks and network mitigation strategies, making it a critical tool for advancing reliable telesurgery across diverse connectivity scenarios.
- 375 trials across 5 network conditions showed success rate drop from 97% (Hospital LAN) to 12% (GEO Satellite)
- Mean task time increased from 80s (Hospital LAN) to 255s (GEO Satellite) for successful trials
- Open-source benchmark uses Linux TC, NetEm, and Gilbert-Elliott model for reproducible evaluation
Why It Matters
Quantifies real-world network limits on telesurgery, guiding design of robust video streaming for remote medical procedures.