Robots are starting to move into the OR. This article presents an overview of how the pioneers in this evolving field are bringing innovative techniques to surgeons.

On the cover: The majority of surgical robots are better described as teleoperators. The best known of these is the da Vinci Surgical System from Intuitive Surgical Inc.
•Defining robotic solutions

•Applicable surgeries

•Inner technologies and components

•Challenges and trends

By Peter Cleaveland, West Coast Editor
Star Trek’s holographic doctor is not yet performing surgery, but robots are finding a place in operating rooms. Some are acting as extensions of the surgeon’s hands and eyes, while others handle the surgeon's scalpels, hemostats, and clamps on voice command. There have been several attempts to build completely autonomous robots that would perform surgery on their own, but so far success in this area has been limited.

This report takes a look at the current state of operating room robotics, examines what goes into a surgical robot, and suggests where designs are headed in the future. It includes information about a novel design that may hold promise for surgical robots in the years to come and an application of small DC motors in designing a benchtop bioreactor.
Extension of Surgeon

The da Vinci surgical robot makes it possible for a surgeon to do things that would be otherwise difficult or impossible.
The majority of surgical robots are better described as teleoperators, systems that allow a human surgeon to do things otherwise difficult or impossible such as complex laparoscopic surgery. The best known of these is the da Vinci Surgical System from Intuitive Surgical Inc. Designed for minimally invasive procedures, it allows a surgeon effectively to put both eyes and both hands inside the patient through a small incision.

Design requirements for such a system are stringent. It must follow the surgeon’s inputs faithfully, never making uncontrolled movements, and it must provide the surgeon with a clear and unambiguous view of what is going on at all times. It must also compensate for any fatigue tremors the surgeon’s hands may develop during the course of a lengthy procedure. There have been several attempts at building true robots for surgery—machines that, like industrial robots, could be programmed in advance and then proceed unaided through an operation. One of the earliest companies to try this was Integrated Surgical Systems, which in 1992 brought out its Robodoc. This robot could be used in primary total hip replacement, milling a cavity in the femur for the placement of a prosthetic implant; in revision hip replacement, removing bone cement and creating a new, accurately shaped cavity for a revision prosthesis; and in total knee replacement, planing knee surfaces on the femur and tibia to achieve a precise fit of the implant. Robodoc was never approved for use in the U.S.

Integrated Surgical Systems’ other robotic product was NeuroMate, an image-guided robotic system for stereotactic functional brain surgeries. It manipulated the operating tools within the surgical field as planned by the surgeon on a separate image planning workstation.Integrated Surgical eventually ran into financial difficulties, and in June of this year laid off all its employees.
Surgical Assistants

The Penelope Surgical Instrument Server robot hands instruments to the surgeon on voice command.
While stand-alone robots aren’t doing actual surgery, they are beginning to move into the operating room in other capacities. In June of this year, the Penelope Surgical Instrument Server, built by Robotic Surgical Tech Inc., was used in its first surgical procedure at New York-Presbyterian Hospital’s Allen Pavilion. Equipped with artificial intelligence (AI), commanded by the surgeon’s voice, and guided by a built-in vision system, it keeps a tray of surgical instruments at the ready and hands them to the surgeon on demand. When the surgeon puts down an implement, the robot notes its presence and decides, based on the context, whether to leave it there or put it back on the tray. “If it thinks that’s going to be the next instrument that he’s going to want, it will just leave it there,” says Michael R. Treat, president and CEO of Robotic Surgical Tech Inc., “because that’s the fastest place to keep it. But if it keeps an eye on an instrument for a while and nothing seems to be happening with it, then it will go and quietly pick it up and put it back.”

The heart of the Penelope system is its software. The upper level, written in Java, runs (at least in the Alpha version used in the first procedure) on a Macintosh laptop computer. “That’s where the image processing and AI is done, and the other tasks the robot has to perform,” says Treat. “It’s where the motion control is done.” The Mac is used at other times for programming the system. Below the Java layer is a middle layer written in C that provides an interface to the lowest-level routines, written in assembler to run on an 8051 microcontroller that does the real-time handling of the motors and the sensors and similar tasks.
Speed, Strength, Precision
Industrial robots are judged on how fast they do their work, and saving several seconds on an assembly line is important. As long as a surgical robot can keep up with the surgeon’s hands, it’s fast enough. While a surgeon may check out the system’s response before starting surgery, says David Rosa, senior director of marketing at Intuitive Surgical Inc., “you move much more slowly than you might think you do when you’re doing surgery.”

An industrial robot may have to exert forces measured in the tens of pounds or more—sometimes much more. Anyone who has visited an industrial trade show that featured robots can remember seeing automobile engine blocks being flung around as if they were toys. The total force required at the end effector is about two pounds. The maximum load for the Penelope system is a bit less than one pound, which is appropriate, says Treat, for the instruments used in general surgery.The degree of precision needed for a surgical robot is a fraction of a millimeter, which is sufficient to do a coronary anastomosis using sutures about the diameter of a human hair.
Choice of Motors
Two types of motors are generally used to drive surgical robots: brush-type DC and brushless DC. How to choose? Intuitive Surgical favors the brush type because of its simplicity. “They’re very well understood,” says Bill Nowlin, the company’s director of software systems, “and so we can build models and run those models against the motors in real time, and check for deviations against the model.” Designing a surgical robot entails enough design risks as it is, he goes on, “and we didn’t want to layer in any more risks than we had to, so we went with conservative design.”

Not everyone uses brush-type motors, however, says Scott Hamilton, sales engineer at Maxon Precision Motors Inc. Some users go for brushless motors because of their long life. A brush motor’s life is limited by its brushes to about 4,000 to 6,000 hours, he says, and that’s if it’s not subjected to heavy overloads or high temperatures. The life of a brushless motor, on the other hand, is determined by its bearings, and can reach 20,000 to 25,000 hours. The price for that is increased complexity in the electronics and the feedback system.
Stability, Feedback, Signal Processing
The choice of feedback device for a surgical robot is crucial. Not only must it provide proper feedback, but it also must be utterly reliable. A common choice for this job is an incremental encoder, often mounted at or near the motor. This gives a good readout on motor position and speed, but since it lacks a zero point there must be a separate feedback device to provide an absolute position reference. In a piece of industrial equipment with incremental encoders—a machine tool, for example—that’s generally done by driving the machine to a reference position at startup to zero the servos. This isn’t such a good choice with a surgical robot, and since safety considerations require redundant feedback paths in any case, the usual procedure is to make the second feedback device an absolute one. “So they either have another linear encoder, or a potentiometer, or even another rotary encoder somewhere else and compare those two signals to make sure they match,” says Hamilton.

Intuitive Surgical’s choice is potentiometers, coupled as closely to the end effector as practical. The system uses the potentiometers to establish initial position but uses the encoders on the motors for feedback when running. Some companies choose absolute encoders for this job, says Hamilton, although “you do need to do a double check occasionally. You need to make sure you haven’t drifted away from where you think you are, especially because sometimes electronics become saturated with information, so you have to shut it down and then re-home it anyway.”

Since a surgical robot generally has a human in the loop, the question arises as to what kind of feedback to provide to that human. Is an endoscope watching the surgical field enough? Many people would say that it is. Intuitive Surgical’s da Vinci system, for example, provides binocular vision scaled to the task at hand so that the surgeon can operate using the same visual clues he would use in conventional surgery. “You make up for the loss of sense of touch with what you’re seeing with your eyes,” says Rosa, “how far tissue is stretching, if it’s blanching, how far you’re pulling your instruments apart if you’re tensioning a suture.”

Despite the success of what Intuitive Surgical calls “visual tactility,” the idea of force feedback is still attractive. The question is how to make this haptic feedback work and how to make it stable. Researchers around the world have been working for some time on haptic feedback systems using things such as strain gages to sense force, but so far we have not seen any in commercial use. The da Vinci system does provide some force feedback, derived from the error signals in the position loops. “I would describe it as a variant of bilateral force reflection,” says Nowlin, “and what that means is that the error produced between desired and actual at the patient-side manipulator is used to generate a feedback signal that the master and, therefore, the surgeon feels.” This is not true force feedback, he points out, for a simple reason: “If that were completely linear and if its scale factor were one, if its gain were one, then that would be true force feedback. It would also violate Bode’s theorem and a few other things and would be unstable.” To preserve stability, the gain is kept down to the point that it’s possible to prove worst-case stability against certain known impedances.
Looking to the Future
How will robots be used in the operating rooms of the future? Rosa sees two main changes. The first will involve the integration of the robot into the OR. Rather than being rolled in and out as needed, it will be integrated into the table or the ceiling, he says, in ways that will streamline the whole process and bring down the cost of doing procedures in general. In other words, “the less OR time you have, the better.”

The other big advance, Rosa continues, will come in the user interface. Today the surgeon must study a CT or MRI image and then look into the viewer on the robot. In the future, the CT and MRI images will be combined with the visual display, similar to a fighter pilot’s heads-up display, to show margins of tumors, important vessels, and the lesion of interest. “And just as the pilot in that platform becomes superhuman if properly integrated with the user interface,” says Nowlin, “so I think a well-designed integration of advanced sensing techniques and advanced data fusion and display and user interface techniques can provide a superhuman experience for the surgeon.”
Many robots are modeled at least somewhat on the human arm with each moveable axis carried by the one below it. This means that each axis carries different mass and has different dynamics, which complicates the control problem. There is one robot platform, however, that features low mass, high stiffness, six degrees of freedom, and the same dynamics on all axes.

Hexapod parallel robots have been tried for surgical applications, but none has succeeded.
Developed in 1994 by Giddings & Lewis and Ingersoll and based on the Stewart Mechanism of 1965 used for flight simulators, it’s called a hexapod and consists of a top and a base linked by six linear actuators, all under computer control. By controlling the length of the legs, the top can be moved as desired. Its advantages include high strength-to-weight ratio, high stiffness, and low inertia. It also has some drawbacks: it has a limited range of motion, the available displacements vary with position, and the fact that all the axes are involved in every move can make control a tricky proposition. “It’s been going on for many years, but it has a lot of problems. It looks simple but it isn’t,” says Stefan Vorndran, director of marketing at Physik-Instrumente-USA, which manufactures hexapod robots for industrial applications.

The idea of a hexapod surgical robot has been tried several times. Several units of Germany’s Fraunhofer-Institut für Produktionstechnik und Automatisierung (IPA) in Stuttgart teamed up a few years ago to create Universal Robot Systems, which developed a hexapod robot called Evolution 1 for spinal surgery in which extreme precision is essential. The heart of the system was a specially modified M-850 hexapod robot built by Physik-Instrumente. This was attached to the arm of a conventional robot, and powered end effectors were in turn mounted to the top platform of the hexapod.

The Evolution 1 received a lot of attention at the time, but to all appearances the effort fizzled out soon after the initial announcements in 2001. Perhaps the most recent work with hexapod surgical robots was done in 2003 by engineers at the Nanyang Technological University in Singapore, working with surgeons at the National Neuroscience Institute. Their hexapod was designed to drill through the temporal during surgery to remove deep-seated brain tumors. Using three-dimensional MRI images, a radiologist and surgeon would map out a path that would avoid important nerves and blood vessels, and then a programmer would write the appropriate instruction to the robot, which would do the job in half the time taken by conventional methods.


A brushless motor is in the aluminum housing at the top of this three-liter glass bioreactor.
Broadley-James Corp. designs and manufacturers sensors, bioreactors, and advanced control systems for bioprocess applications in the life sciences industries. When the company set out to create a line of glass benchtop bioreactors that were compact and different from others on the market, they had some specific motion control needs.

The bioreactors were to be used with the company’s latest BioNet Bioreactor Control System for both R&D and process development work. Small bioreactors are used by biopharmaceutical companies to create novel proteins with therapeutic properties. A critical parameter for the devices is agitation, which must be closely controlled and monitored in order for the bioreactor to yield reproducible results.

Broadley-James Corp. needed to find a small brushless motor that could be used to provide motion in the agitator for the glass benchtop bioreactors most often used for cell-culture and microbial fermentation. The company preferred brushless motors for their low maintenance requirements and long life. The motors needed to be small enough to fit into a custom motor housing at the top of the bioreactor and couple to the agitation shaft. Whenever the device underwent an autoclave procedure, the motor needed to be removed. As a result, the entire motor and housing assembly had to be easily detachable. And even though the motor had to be compact in size, it still had to deliver high torque over a wide RPM range.

Maxon 40-mm brushless motors fit the design team’s criteria. With their precision encoders, they allowed Broadley-James Corp. to precisely measure the agitator shaft RPM and control the agitation over a wide range of motor speeds. According to Scott Broadley, the company’s president, “the motors are made to tight tolerances and we had no problem quickly designing a housing that the motors fit into interchangeably.”

The brushless motors are electronically commutated, thus allowing extremely long motor life since there are no mechanical brushes to wear out. Using electronic commutation also minimizes electrical noise. High-energy neodymium magnets make for a very responsive motor while minimizing overall size. The encoder provides both commutation signals and a three-channel signal to provide feedback to the control electronics. Key motor requirements included a precision encoder for reproducible management of the bioreactor’s agitation rate. The motor’s continuous output power is about 200 watts, maximum speed is 12,000 RPM, and maximum continuous torque is up to 265 mNm (37 oz-in.) depending on the winding. The motor has an ambient temperature range of –20 to +100°C (–4 to +212°F) and is built to IP54 standards.
For additional information on the technologies and products discussed in this article, see Medical Design Technology online at and the following websites: