Applying a Model-Based Approach for the Development of Microsurgical Robots

Microsurgical techniques and tools enable surgeons to perform intricate procedures on nerves, blood vessels, and tiny structures of human tissue with exceptional precision. While microsurgery shows great promise for improved patient outcomes, operating within tightly constrained workspaces at submillimeter scales requires extraordinary steadiness and dexterity. A skilled surgeon can connect vessels between 0.3 and 0.8 millimeters (mm) in diameter. However, the inevitable physiological tremor of the surgeon’s hand limits the efficacy of such anastomosis procedures. Additionally, workspace and kinematic challenges in a confined anatomy can require multiple design iterations that are both time-consuming and costly.

Microsurgical robots can help surgeons overcome limitations imposed by narrow workspaces, hand tremors, and fatigue. The design of these robots, however, introduces a new set of challenges. First, to reduce the time that surgeons spend learning new tools, the robots must be capable of assisting with different types of procedures, not just one. Second, unlike industrial robots used in workspaces with no humans nearby, microsurgical robots are used directly on humans, and thus must be designed with patient safety as a top priority. Finally, to minimize tissue damage and reduce recovery times, the robots must be minimally invasive—ideally working through a single, small incision.

Many design decisions hinge on minimizing intrusiveness while ensuring the surgeon has sufficient degrees of freedom to perform procedures effectively. There is no one-size-fits-all answer to this tradeoff question, and, as a result, microsurgical robotic design teams have often relied heavily on trial-and-error approaches. They must specify requirements, create a design to meet those requirements, and then assemble a prototype. Design teams then test the prototype to further refine requirements before repeating this cycle. Often, multiple iterations of this cycle are needed, and cycle times are slowed by the need to build or rebuild the hardware prototype on each iteration.

At Zhejiang University, my colleagues and I have applied a design-centric, model-based approach to the development of systems for robot-assisted minimally invasive surgery. Using this approach, we recently designed a robotic manipulator for anastomosis and ophthalmic procedures based on a parallelogram structure. We performed computational analysis and simulation tests with Simulink^® and Simscape Multibody™ to visualize the manipulator’s end-effector trajectory and confirm that the design meets requirements for safety and surgical operability (Figure 1). This approach speeds development and enables us to identify and resolve many design issues before investing time and resources into building a hardware prototype.

Designing the Mechanical Structure

We began our design process by defining requirements and design goals for the microsurgical manipulator. These included, for example, an accuracy target of less than 10 micrometers for the tip, a motion range of 20x20x20 mm, and a quick-change mechanism for the system’s end-effector that will enable surgeons to rapidly replace instruments during a procedure.

A key design component of the system is the remote center of motion (RCM) mechanism, which restricts the degrees of freedom (DOF) of the instrument to three rotational DOF (ψ, ϕ, and θ) around the incision and one translational DOF (Z) in the direction of instrument insertion. We designed a double-parallelogram structure that enables movement of the end-effector throughout the workspace with the following motion ranges: ψ: ±45°；ϕ: ±75°；θ: 360°; Z: 32 mm. After first analyzing this structure via a mathematical model based on first principles, we created a CAD assembly for it in SolidWorks^® (Figure 2).

Performing Simulation-Based Analysis in Simscape Multibody

Our next step was to export the CAD assembly from SolidWorks using the Simscape Multibody Link plug-in, and then import the resulting XML multibody description file into Simscape Multibody to create a Simscape™ model of our design (Figure 3). We added motors at the ϕ, ψ, and Z joints for motion control and ran multiple simulations using a position sensor to track the position and motion of the end-effector.

By analyzing and plotting the results of these simulations in MATLAB^®, we visualized the range of the end-effector within the cubic space for anastomosis procedures and within the spherical space for ophthalmic procedures (Figure 4). This is to ensure safety for the patient and a high probability of surgical success for the patient, with all points in the anatomy reachable during a procedure. This assessment requires using the robot’s kinematics to develop a point cloud and evaluate trajectories seen in traditional surgical interventions.

Figure 4. Multiple views of the workspace for ophthalmic procedures (first) and anastomosis procedures (second).

We also ran simulations in which we tracked the movement and trajectory of the end-effector, for example, to confirm that the design met the surgical requirement for an arcing motion on the surface of the eye (Figure 5).

Key Advantages of a Model-Based Approach

In our research, we see several key advantages to the model-based approach we have adopted. Among the most valuable is the ability to rapidly iterate the design to make sure it is fully functional in the digital realm before moving to a physical implementation. It’s one thing to create a design on paper or in CAD software and know that it should work in theory, but, when we see it working in a virtual environment via simulation with Simulink and Simscape, we inevitably gain further insights to improve it.

With the digital models readily available in the model-based approach, we can further shorten the development lead time by using 3D printing technology for rapid prototyping of the robot mechanism in combination with the simulation analysis. The hardware testing of the design derived from the model-based approach shows that the prototype successfully maintains the RCM point on the patient’s eye (Figure 6).

In addition, working in Simulink and Simscape makes it easier for the many students who work on our team to come up to speed and coordinate their efforts. Students move from project to project before they graduate, and a model-based approach helps them pass on what they have done to others. Models are easier to explain and to understand, so when the next group wants to extend or generalize the work of the previous team, they know where to start and what to do.

Lastly, now that we have demonstrated the feasibility of the mechanical design, we are ready to begin development of the control system and, as we do so, further optimize the structure. Model-Based Design with MATLAB and Simulink will help with both of these endeavors, enabling us to validate the safety of the complete system via simulation and shorten the overall development cycle.