Researchers at Aalto University in Finland are developing a surgical navigation system that uses a 3D camera to guide robots during knee replacement operations, potentially eliminating the need for reflective markers that are currently screwed into patients' bones, according to a university news release.
The ORCA research group, which started the project last year, aims to create a system that tracks bone movement automatically using machine vision, removing what the team describes as a fiddly and time-consuming calibration process that can add roughly 15 minutes to an operation.
Around 3.5 million hip or knee replacement surgeries are performed each year in OECD countries alone — more than 5,000 knee replacements daily, according to Aalto. A typical knee replacement takes just over an hour. While roughly 90 percent of patients are satisfied with the outcome, the research group says its primary goal is not to improve precision further but to simplify the procedure.
"We want to eliminate all the extra steps and equipment that take time, increase risks, and ultimately add costs," Ville Pitkänen, a researcher on the ORCA project, told Aalto. "Our vision is that the navigation system will start tracking the bone automatically and help the surgeon as soon as the operation begins."
Pitkänen first wondered why 3D cameras weren't used in joint replacement surgery nearly 20 years ago while working on his master's thesis, which touched on computer-assisted orthopaedic surgery. He later worked for a company supplying robots for joint replacement operations before joining the ORCA project.
The problem with markers
In current robotic-assisted knee replacement surgery, surgeons screw two or three palm-sized reflective markers into the patient's bones. These cage-like structures, fitted with spheres that reflect infrared light from a camera, allow the robot to know where the bones are and how they move during the procedure.
The surgeon must then calibrate the system by touching the bone at several dozen points with a sharp-tipped tool that also carries a marker. If the markers shift during the operation, the entire calibration must be repeated.
Pitkänen, who has followed hundreds of operations, told Aalto that using markers can easily add about 15 minutes to an operation, even though the bone work itself goes faster with the robot's help. He noted a link between longer operations and increased infection risk. "The risk of infection also increases when extra instruments are used and extra holes are drilled into the bone. Some of the markers are also attached through the skin," he said.
The ORCA system replaces those markers with a 3D camera that measures the bone surface directly. A machine vision application identifies the bone from the point cloud generated by the camera — tens of thousands of data points — and determines its position. The current marker-based system tracks just 16 points (four markers with four spheres each).
"Our navigation system can distinguish bone from other tissue," Pitkänen said.
Testing and accuracy
The researchers have optimised the camera to make the system accurate enough for clinical use and filed their first patent applications in October 2026, according to Aalto. In tests, the team connected their system to a GoFa robot manufactured by ABB, a type commonly used in the electronics industry. When Pitkänen attached a laser to the robotic arm, it pointed at exactly the same spot on a plastic bone every time, even when that spot was not in the camera's view. Moving the bone, the laser continued tracking the invisible point.
"We can see a certain part of the bone, and then we know where the whole bone is. We don't need to see the spot where the drilling will take place," Pitkänen explained.
The team has also tested the system on animal bones — a whole pig's leg supplied by food company Atria — and plans to move to testing on cadavers donated for medical research, in collaboration with doctors. They also intend to observe joint replacement operations in the operating theatre.
Growing use of surgical robots
Robots are currently used in around 10–20 percent of knee replacement operations worldwide, according to Pitkänen. Finland first used them in 2021. Australia has been a pioneer, with robots used in 42 percent of knee operations in 2024. Pitkänen expects these figures to grow.
He drew an analogy with car technology: "Quite a lot of people use cruise control and lane assist in their cars, even though they could get to their destination without them."
As robot use increases, so does the need for an easy-to-use navigation system, he said. While greater precision is not the main goal, replacing reflective markers with a 3D camera will improve it, he added.
From hospital observation to commercialisation
The idea for the 3D camera system emerged partly from the team's participation in the Biodesign Finland programme, developed at Stanford University, which brings together multidisciplinary teams to identify healthcare problems and devise commercialisable solutions. Aalto and HUS, the Helsinki University Hospital, host the programme together each year.
"We spent a few months walking around HUS's Siltasairaala hospital looking for problems and coming up with solutions," Pitkänen said.
The ORCA project secured Business Finland Research to Business funding, which aims to help prepare research results for commercialisation. The team is exploring options including licensing the technology or setting up a start-up.
Pitkänen believes the market is large because the navigation system could be connected to the control system of any robot. It could also serve as an extra set of eyes in joint replacement operations where the surgeon is not using a robot. "An easy-to-use system could make computer-assisted navigation an attractive option for hospitals where buying an expensive robot does not make sense. It might cost a tenth of the price of a robot," he said.
Once the 3D camera has proven itself in knee replacement surgery, the research group plans to move on to other types of joint replacement and bone surgery, including spinal and brain surgery. "All of them use reflective markers. We want to replace those with a 3D camera," Pitkänen said.
Why machine vision hasn't been used before
The ORCA research group has a few explanations for why machine vision-based navigation has not become established in joint replacement surgery. Sauli Lindroos, who focuses on biomedical imaging, told Aalto that computing power may only now be reaching a sufficient level, as tracking the tens of thousands of points produced by a 3D camera requires significant computation.
Machine vision algorithms have also developed to the point where objects can be recognised and located more quickly and accurately than just a few years ago, making it possible to distinguish bone from surrounding tissue.
The research group is careful not to describe their system as artificial intelligence. Taneli Tikkakoski, whose background is in bioinformatics, told Aalto that could bring to mind generative AI, which can produce different answers from the same input — the opposite of what ORCA aims for. "We have chosen established algorithms that are predictable and repeatable. When given the same input, they always produce the same result," he said.
The result is machine vision that is simple and precise. It does not think for the surgeon or learn on the fly. It simply sees and measures accurately.