Robotic joint replacement

• The robot makes no decisions: it keeps the instrument within the surgeon's plan.
• Placement precision is higher, which is proven. No difference in pain or function at one year was found (RACER-Knee).
• Self-pay starts at 650,000 roubles; some state centres use the robot under a VMP quota.
• The surgeon's experience matters as much as the technology. Choosing a system often means choosing an implant brand.

The robot and the surgeon

In joint replacement the robot is a computer-controlled tool. Clinics write that the robot operates, yet it makes no clinical decisions.

The surgeon does the essential work: examines you, chooses the operation and implant, makes the incision, prepares the bone and fits the implant. In most systems the robot has one job: keeping the instrument within a plan made in advance. Active systems such as CUVIS-joint mill the bone themselves, under the surgeon’s constant control[1]1Robotic-assisted total knee arthroplasty: precision versus patient-level outcomes — critical review — PMC, 2026To the list of sources →.

The computer knows where the bone cut should go. If the surgeon’s hand drifts, the instrument slows or stops. This removes random errors; it does not replace the surgeon’s experience.

The surgeon guides the saw. The system limits its movement to the plan.

A simplified illustration of a system in which a robotic arm limits how the saw can move. Not every robot works this way.

Skip the animation

Illustration

Markers for the cameraVirtual boundaryA retractor protects soft tissueThe surgeon guides the saw on the robotic armFlat cut surfaceFurther passes and lifting out the layer are not shown

Step 1 of 7

Plan. Using scans, the surgeon plans the implant position and the level of the cut.

← BackNext →

All 7 steps

1. 1Plan. Using scans, the surgeon plans the implant position and the level of the cut.
2. 2Bone check. The surgeon matches the bone to the model with a probe, then checks the ligaments and refines the plan.
3. 3Boundary. From the plan the surgeon approved, the system sets a boundary for the saw.
4. 4Saw approach. The surgeon brings in the saw on the robotic arm. Retractors hold the ligaments aside.
5. 5The cut. The surgeon guides the saw in several passes within the cutting plane.
6. 6The stop. At the plan boundary, the system limits how the saw can move. The surgeon protects ligaments and soft tissues.
7. 7Checking. After the layer is removed, the surgeon checks the cut. Further passes and lifting out the layer are not shown.

1. 1Plan. Using scans, the surgeon plans the implant position and the level of the cut.
2. 2Bone check. The surgeon matches the bone to the model with a probe, then checks the ligaments and refines the plan.
3. 3Boundary. From the plan the surgeon approved, the system sets a boundary for the saw.
4. 4Saw approach. The surgeon brings in the saw on the robotic arm. Retractors hold the ligaments aside.
5. 5The cut. The surgeon guides the saw in several passes within the cutting plane.
6. 6The stop. At the plan boundary, the system limits how the saw can move. The surgeon protects ligaments and soft tissues.
7. 7Checking. After the layer is removed, the surgeon checks the cut. Further passes and lifting out the layer are not shown.

An illustration, not a photo of surgery. The skin incision, the hands of the surgeon and assistant, sterile covers, the camera and the robot itself are not shown, only the end of the arm. The thigh bone, the ligaments and, during the cut, the top layer of bone are shown see-through, and the camera markers are drawn smaller; lifting out the cut layer of bone is not shown. The steps are shortened: the saw makes more passes in reality, matching the bone to the model and checking the ligaments take longer, and surgeons cut in different orders. This version of the operation removes the cruciate ligaments. A screen cannot convey force, so the stop is shown with movement and words.

The robot helps carry out the implant plan more precisely. In the RACER-Knee trial (339 patients), no additional improvement in pain or function over conventional surgery was found at one year. There is no evidence yet that implants placed with a robot last longer.

Watch from the start

The operation step by step

For you, a robotic operation looks almost the same as a standard one. The anaesthesia, the incision and the implant are the same[4]4Total Knee Replacement — AAOS OrthoInfo, 2024To the list of sources →. The preparation and a few steps in the operating room differ.

In the RACER-Knee trial the robotic operation took about 10 minutes longer on average[16]16Robotic-assisted knee replacement no better for patients than conventional surgery (RACER-Knee press release) — University of Warwick, 2026To the list of sources →. While the surgeon is learning the system, the difference can be greater[1]1Robotic-assisted total knee arthroplasty: precision versus patient-level outcomes — critical review — PMC, 2026To the list of sources →. Recovery is the same as after standard surgery: walking with support in the first days, discharge after a few days[4]4Total Knee Replacement — AAOS OrthoInfo, 2024To the list of sources →.

Classes of systems

Robots for joint replacement differ in how they work. We group them into four classes, plus augmented-reality navigation. NICE has included five systems in its UK evidence-generation programme: Mako, ROSA, VELYS, CORI and SkyWalker[3]3Robot-assisted surgery for soft tissue procedures and orthopaedic procedures (HTE22) — NICE, 2025To the list of sources →.

The mark for Russia means only that we confirmed the system from open sources. It says nothing about official registration.

Placement precision

The robot consistently reduces how far the implant sits from the plan. Outliers, implants placed noticeably off target, are fewer with a robot[1]1Robotic-assisted total knee arthroplasty: precision versus patient-level outcomes — critical review — PMC, 2026To the list of sources →. For the hip, a meta-analysis found more accurate cup placement[5]5Robotic versus conventional total hip arthroplasty: systematic review and meta-analysis — The Journal of Arthroplasty (PMID 39710214), 2025To the list of sources →. The result repeats across studies of different quality and is hardly disputed.

Pain and function at one year

In September 2026 The Lancet published the results of RACER-Knee[2]2Robotic-arm assisted versus conventional total knee replacement (RACER-Knee): a randomised controlled trial — The Lancet, 2026To the list of sources →. It is the largest blinded randomised trial of knee replacement with the Mako robot. Patients did not know whether a robot was used. The trial included 339 people[2]2Robotic-arm assisted versus conventional total knee replacement (RACER-Knee): a randomised controlled trial — The Lancet, 2026To the list of sources →.

The main outcome was the Forgotten Joint Score (FJS-12). It measures from 0 to 100 how far a person stops noticing the artificial joint. At 12 months the score was 49.2 with the robot and 50.2 without. The adjusted difference was −1.5 points, p = 0.62[2]2Robotic-arm assisted versus conventional total knee replacement (RACER-Knee): a randomised controlled trial — The Lancet, 2026To the list of sources →. No difference was found on the main outcome. The robotic operation cost more[2]2Robotic-arm assisted versus conventional total knee replacement (RACER-Knee): a randomised controlled trial — The Lancet, 2026To the list of sources →.

For the hip the picture is similar. A 2024 meta-analysis confirmed more accurate placement and found no advantage in clinical scores[5]5Robotic versus conventional total hip arthroplasty: systematic review and meta-analysis — The Journal of Arthroplasty (PMID 39710214), 2025To the list of sources →.

NICE, the British institute, assesses medical technologies for the public health service. In 2025 it allowed robots only while new evidence is collected[3]3Robot-assisted surgery for soft tissue procedures and orthopaedic procedures (HTE22) — NICE, 2025To the list of sources →.

Implant lifespan

Standard implants last a long time: at 25 years about 82 % of knee[6]6How long does a knee replacement last? A systematic review and meta-analysis of case series and national registry reports with more than 15 years of follow-up — The Lancet, 393(10172):655–663, 2019To the list of sources → and about 58 % of hip implants are still working[7]7How long does a hip replacement last? A systematic review and meta-analysis of case series and national registry reports with more than 15 years of follow-up — The Lancet, 393(10172):647–654, 2019To the list of sources →. The younger the patient, the more likely a repeat operation will be needed one day[7]7How long does a hip replacement last? A systematic review and meta-analysis of case series and national registry reports with more than 15 years of follow-up — The Lancet, 393(10172):647–654, 2019To the list of sources →.

Robots have been used widely for less than ten years, so long-term data are scarce[1]1Robotic-assisted total knee arthroplasty: precision versus patient-level outcomes — critical review — PMC, 2026To the list of sources →. There is no evidence that a robot reduces repeat operations[1]1Robotic-assisted total knee arthroplasty: precision versus patient-level outcomes — critical review — PMC, 2026To the list of sources →. Precision may help in theory, yet this remains a hypothesis.

General risks of surgery

The robot does not remove the general risks of joint replacement. Implant infection in the first year after a primary operation occurs in about 1.2–1.3 % of cases[8]8Periprosthetic joint infection after 1.66 million primary and revision arthroplasties (2015–2023) — PMC, 2025To the list of sources →. After repeat operations it is more common, at 6.5–8.8 %[8]8Periprosthetic joint infection after 1.66 million primary and revision arthroplasties (2015–2023) — PMC, 2025To the list of sources →. Russian authors report infection in about 3 % of patients[9]9Эпидемиология эндопротезирования тазобедренного и коленного суставов и перипротезной инфекции в Российской Федерации — Травматология и ортопедия России, 27(3):84–93, 2021To the list of sources →.

Blood clots, dislocation, unequal leg length and stiffness are possible with a robot too. Recovery, restrictions and warning signs are the same. They are covered on the knee and hip pages.

Clinics with robots

The list shows clinics whose systems we confirmed from open sources. It is incomplete. Check the system, quota options and price with the clinic.

Cost

Under a VMP quota. The state pays for high-tech care, and the robot is not a separate line item[10]10Травматология и ортопедия по квоте (ВМП): как получить — gosmed.ru, 2025To the list of sources →. If a surgeon at a state centre uses the system within a quota, it is free for you. You cannot choose the robot on request. The wait for a quota ranges from weeks to months[10]10Травматология и ортопедия по квоте (ВМП): как получить — gosmed.ru, 2025To the list of sources →. How to get one is explained on the free surgery page.

Self-pay. Clinics’ published prices for a robotic operation range from 650,000[11]11Эндопротезирование коленного сустава роботом — Д. А. Чугаев (chugaev.info), 2026To the list of sources → to about 870,000 roubles[12]12Эндопротезирование роботом-хирургом — ГК «Мать и дитя», Клинический госпиталь Лапино, 2026To the list of sources →. A standard self-paid replacement in the same cities is noticeably cheaper. The difference goes on consumables, CT planning and system maintenance.

A clinic may quote a price without CT and consumables. At one clinic on the list they add more than 170,000 roubles to the operation[12]12Эндопротезирование роботом-хирургом — ГК «Мать и дитя», Клинический госпиталь Лапино, 2026To the list of sources →. Ask for a full estimate. The outcome of surgery is individual, and promising a result is not allowed[13]13Федеральный закон № 38-ФЗ «О рекламе», ст. 24 — КонсультантПлюс, 2006To the list of sources →.

Who might benefit from a robot

Research does not identify a group that must have a robot. There are situations, though, where precise placement matters most. These are topics to raise with your surgeon, not a recommendation.

• Marked joint deformity. When the leg axis is badly bent, a 3D plan helps work out the cuts in advance.
• Hip dysplasia. With a socket misshapen from birth, precise cup placement may matter[5]5Robotic versus conventional total hip arthroplasty: systematic review and meta-analysis — The Journal of Arthroplasty (PMID 39710214), 2025To the list of sources →.
• Partial knee replacement. Partial knee implants are revised more often than total ones[15]15Adverse outcomes after total and unicompartmental knee replacement in 101 330 matched patients: a study of data from the National Joint Registry for England and Wales — The Lancet, 2014To the list of sources →. This is where robots have the longest track record.
• Revision operations. Robotic protocols for revisions exist, yet data are still limited.
• A surgeon early in the learning curve. The system helps keep results consistent[1]1Robotic-assisted total knee arthroplasty: precision versus patient-level outcomes — critical review — PMC, 2026To the list of sources →.

A robot adds least for a patient with standard anatomy and an experienced high-volume surgeon. Then choose the surgeon first.

It is a good sign if the surgeon talks calmly about the robot’s limits. Be wary of promises of surgery without pain or an implant for life: no system can offer that.