Powering Next-Generation Surgical Robotics

Surgical robots demand extreme precision. When a robotic arm assists a surgeon in placing an orthopedic implant, a millimeter of error can mean the difference between a successful procedure and a complication that extends recovery time or requires revision surgery.

The Run-Time Reconfiguration Challenge

Surgical robotics requires adaptability. Procedures vary—different tools, layouts, and teams create a dynamic environment where a robot must operate safely among people.

But beyond adapting from one procedure to the next, MinMaxMedical needed the robot to adapt during a procedure as surgeons change tools or as the environment shifts.

“The ability to modify robot models at run time has significantly enhanced our workflow,” says Cédric Beausse, the business unit director for surgical robotics at MinMaxMedical. The team can now test new tools and configurations without stopping to regenerate code.

The technical innovation centers on Robotics System Toolbox and its rigidBodyTree data structure, which represents the kinematic chain of a robot manipulator. Traditionally, this structure would be defined once and remain static. MinMaxMedical worked with MathWorks to enable run-time modifications, allowing the software to add or remove not just additional bodies, but also add entire trees of tools to the main platform’s tree while the system is running. This capability transforms the robot from a fixed machine into a reconfigurable platform.

Collision avoidance adds another layer of complexity. The robot must check that its planned motions won’t cause it to hit the patient, the surgical team, equipment, or itself. MinMaxMedical’s system can incorporate new collision objects at run time, including meshes and point clouds from depth sensors.

“The number and the geometry of collision bodies can change depending on what you want to do at run time,” Beausse says.

These checks are computationally expensive, and they must complete fast enough to maintain real-time control. A collision check that takes too long could allow the robot to move into a dangerous position before the system catches the error. The collision detection algorithm includes parameters that let engineers balance precision against speed based on each application's requirements.

The practical implications extend beyond MinMaxMedical’s own products. The company provides robotic components and software to other medical device startups through its incubator. The ability to reconfigure at run time means these clients can adapt the platform to their specific surgical applications without requiring custom code generation for each use case.

Building the System Before the Hardware

Kinematics and collision avoidance don’t run themselves. A surgical robot needs supervisory logic to coordinate its motion controllers, safety systems, and tool management, determining what to do at any given moment.

MinMaxMedical built this in Stateflow. Rather than writing complex nested if-then-else logic in code, engineers draw diagrams showing the different states the system can be in and the transitions between them. “Stateflow is the brain of our controller,” Beausse says. “It commands every part of our software and is visual enough to be understood by everyone on the team, including non-engineers.”

This visual accessibility is part of a broader approach. Model-Based Design treats simulation models as the central artifact of the development process rather than code. Engineers build and test algorithms in Simulink, verifying behavior across a wide range of simulated conditions before any code is generated. Once the models are validated, Embedded Coder and Simulink Coder™ automatically generate production code.

“Model-Based Design helps us try out features without having to go to the hardware,” Beausse says. “It saves time but also shows us bugs that could make the robot dangerous to itself.” Engineers can simulate sensor failures, communication dropouts, unexpected obstacles, and other abnormal conditions that would be difficult or dangerous to reproduce with a physical robot.

Medical devices sold in Europe and the United States must comply with IEC 62304, which requires extensive documentation, traceability, and verification. Meeting these requirements through traditional development can consume significant engineering time. When the model is the source of truth, documentation and traceability come along for free. “Model-Based Design saved us a lot of time for our QA process, which is bringing us to 62304 compliance,” Beausse says.

Not everything can be simulated. Some hardware arrives from suppliers as a black box, with behavior that can only be characterized through testing on the physical device. When these components don’t behave as expected, debugging gets difficult.

MinMaxMedical ran into this. “We had some black-box hardware that we couldn’t simulate in Simulink,” Beausse says. Simulink Real-Time™ let the team run control algorithms on target hardware while maintaining visibility into the system’s internal states. When something went wrong, they could see exactly what the software was doing at the moment of failure. The tool also guaranteed real-time execution and handled bus communication automatically.

Scaling the Technology

“Now that we have a good grasp of the tools, we could port our models to other kinematics and architectures,” Beausse says. A robot designed for orthopedic surgery might have very different physical dimensions than one designed for neurosurgery, but the fundamental control principles remain the same.

The company’s incubator model continues alongside manufacturing. MinMaxMedical aims to help create 100 startups by 2035 in partnership with investors and physicians. These ventures would represent more than 10,000 jobs, according to company projections, with many building on the robotic platform MinMaxMedical now manufactures.

“The product was developed in record time thanks to prototyping tools and Model-Based Design,” Beausse says.

Additionally, this type of run-time model-based flexibility is also becoming important in industrial robotics, making the approach scalable beyond medical devices. Industries such as automotive and electronics manufacturing increasingly need robots capable of switching between tasks, tools, and layouts throughout the day. The same run-time reconfigurable model could allow a single robot to handle multiple operations, improving throughput and reducing the need for task-specific hardware.

Lessons for Safety-Critical Robotics

MinMaxMedical’s experience offers insights to any engineering team developing safety-critical robotic systems. The challenges it faced—including run-time reconfiguration, real-time collision avoidance, regulatory compliance, and integration of black-box components—arise across industries from medical devices to autonomous vehicles to industrial automation.

Staff collaborating on future technologies. (Image credit: MinMaxMedical)

Building a system at the edge of what’s possible sometimes means pushing the tools forward too. “MathWorks was responsive to the issues we had, especially with Robotics System Toolbox,” Beausse says. “We were able to share our input, and they helped precisely with what we thought should be the focus of it.” That collaboration produced improvements to the toolbox that benefit not just MinMaxMedical but any customer building reconfigurable robotic systems.

The surgical robotics market continues to grow rapidly, driven by aging populations, demand for less invasive procedures, and the potential for robotic assistance to improve surgical outcomes. MinMaxMedical’s new factory positions the company to capture a share of that growth, with robots that combine precision engineering, intelligent software, and the flexibility to adapt to the varied demands of modern operating rooms. The technical foundation built through Model-Based Design and close collaboration with MathWorks will support that ambition as the company scales from a startup to an industrial manufacturer.