Robotic Lifestyle

Telesurgery's next act: what it will take for humanoids to work in the operating room illustration
Humanoids·4 min read

Telesurgery's next act: what it will take for humanoids to work in the operating room

By Sophia Chen

Surgeons in one city manipulating robots hundreds of miles away is no longer a research vignette; it is a commercial roadmap. Funding, foundation models and digital twins are aligning to make remote-enabled surgical platforms viable - but humanoid helpers in the OR face a different, steeper engineering climb.

The momentum is measurable. Sovato Health, a middleware company for remote procedures, says more robots will ship remote-enabled in the next 12 months and is already supporting device makers through regulatory filings; Physical Intelligence closed a $600 million Series B to train robots on real-world contact tasks; Parallax Worlds raised seed capital to stress-test robots in hyper-realistic digital twins. Together, these developments compress the R&D timeline for surgical robotics and for broader, physically capable robots in clinical settings.

Why telesurgery is pushing hardware and software to meet

Yet the path from a telesurgery-enabled arm to a humanoid that can scrub, reposition instruments and respond to complications is long. The OR demands hard real-time control, force-sensing fidelity, predictable failure modes and explicit safety cases for every degree of freedom. Engineers and hospitals are racing to build that stack while regulators insist on evidence - meaning pilots, 510(k) dossiers and interoperable cybersecurity before widespread adoption.

Telesurgery's immediate commercial case is straightforward: health systems want to extend scarce specialist time across geography and increase operating-room utilization. According to market analysis cited by The Robot Report, the U.S. performs roughly 10 million major operating-room procedures annually, yet only about 2.5 percent are robotic-assisted today, leaving ample headroom for growth.

Foundation models and digital twins accelerate physical learning

That opportunity explains why device developers are designing systems to be “telesurgery-native,” Sovato co-founder and CEO Cynthia Perazzo told Medical Design & Outsourcing. Sovato, which closed a $26 million Series B that included investment from Intuitive Surgical, positions itself between device firms and health systems to supply the connectivity, latency management and operational workflows hospitals will need.

The requirements drive specific engineering work: deterministic video and command transport, encryption and auditability, and safety envelopes that preserve force and motion limits when a surgeon is remote. Sovato’s team has co-authored technical guidelines with cybersecurity firm Medcrypt and device teams from Medtronic, Johnson & Johnson and Intuitive, the company says; those documents form the kind of interface and test specifications regulators will expect during 510(k) or equivalent submissions.

Engineering gaps: force, latency, and predictable failure modes

Two software trends are shortening the timeline from prototype to clinical pilot. First, foundation models for physical tasks let robots generalize manipulation and perception with less task-specific engineering. Physical Intelligence says its vision-language-action model can predict 50-step motion sequences in about 100 milliseconds and that its RECAP training reduced failure rates while doubling throughput on complex assembly tasks.

Second, high-fidelity digital twins let engineers stress-test edge cases before metal and motors see them. Parallax Worlds converts simple video walkthroughs into interactive 3D environments that run a robot’s actual control software, the company says. Investors and early customers argue that testing perception, planning and edge-case handling in simulation compresses on-site iterations that used to take months into weeks.

Together these tools change what it means to raise a robot’s TRL, or technology readiness level. Where traditional development required repeated hardware iterations at increasing scale, teams can now train policy and perception in simulation and then validate behavior on a small set of physical trials. That reduces cost and - crucially for hospitals - yields measurable reliability metrics before clinical pilots.

Who wins and what real hospital pilots will look like

Engineering gaps: force, latency, and predictable failure modes

A surgical arm built for precise, repeatable incisions is not the same engineering problem as a humanoid moving through an OR. Humanoids introduce articulation, locomotion and multi-contact interactions that amplify risk. Every additional actuator needs a defined safety envelope and a failure mode the clinical team understands and can mitigate.

Force sensing and haptic fidelity are especially thorny. Telesurgery relies on a surgeon perceiving tissue behavior through instrument responses; humanoid assistants will need comparable force control to perform tasks such as retraction or sponge exchange without causing harm. Those capabilities demand high-bandwidth sensing, low-latency control loops and certified limit-checking hardware.

Cybersecurity and interoperability are a second axis of risk. Sovato’s model is device-agnostic, but that requires standardized handshakes between robot firmware, hospital networks and remote-control middleware. Regulatory reviewers will expect demonstrable protections - authenticated endpoints, tamper-evident logs and fail-safe defaults - as part of any 510(k) or CE submission.

Sources