ORTHOROBOTIC® ARM - A versatile lightweight robot for surgical applications

Abstract

Purpose – Surgical robotics can be divided into two groups: specialized and versatile systems. Versatile systems can be used in different surgical applications, control architectures and operating room set-ups, but often still based on the adaptation of industrial robots. Space consumption, safety and adequacy of industrial robots in the unstructured and crowded environment of an operating room and in close human robot interaction are at least questionable. The purpose of this paper is to describe the ORTHOROBOTICS ARM, a new versatile lightweight robot for surgical applications.

Design/methodology/approach – The design approach of the ORTHOROBOTICS ARM robot focuses on compact, slim and lightweight design to assist the surgeon directly at the operating table without interference. Significantly reduced accelerated masses (total weight 10 kg) enhance the safety of the system during close interaction with patient and user. Additionally, ORTHOROBOTICS ARM integrates torque-sensing capabilities to enable close interaction with human beings in unstructured environments.

Findings – A payload of 30 N, optimized kinematics and workspace for surgery enable a broad range of possible applications. Offering position, torque and impedance control on Cartesian and joint level, the robot can be integrated easily into telepresence (e.g. endoscopic surgery), autonomous or soft robotics applications, with one or multiple arms.

Originality/value – This paper considers lightweight and compact design as important design issues in robotic assistance systems for surgery. Keywords Robotics, Surgery, Kinematics, Product design

1. Introduction Surgical robotic systems can be divided into two major groups: specialized and versatile systems. Specialized systems focus either on a dedicated surgical technique, like endoscopic surgery with the da Vinci surgical system by Intuitive Surgical (Green et al., 1995) and Artemis (Schurr et al., 1996) or on the treatment of a specific medical disease (e.g. cancer in Phee et al., 2005). These systems can fulfil the dedicated task very well, but link the financial amortization in the clinic to single medical procedures. With ongoing research in medical treatment, many of these specialized robotic systems are likely to loose their niche. On the other hand, today’s versatile systems often still base on the adaptation of industrial robots (e.g. Caspar in Albers et al., 2007). Industrial robots are targeted on high-absolute accuracy which is achieved by stiff structures and thus relatively high mass. Safety and adequacy in the unstructured and crowded environment of an operating room combined with close human robot interaction are therefore at least questionable. In contrast, the design approach of the KINEMEDIC and the new generation ORTHOROBOTICS ARM aims at a compact, slim and lightweight robot (LWR) arm as a versatile core component for various existing and future medical robotic procedures. With its low weight of 10 kg and dimensions similar to those of the human arm, the ORTHOROBOTICS ARM robot can assist the surgeon directly at the operating table where space is sparse. Like the LWR (Albu-Scha¨ffer et al., 2007), ORTHOROBOTICS ARM integrates torque sensing capabilities to enable close interaction with humans in unstructured environments. Accelerated masses are reduced by consequently employing the lightweight approach for the ORTHOROBOTICS ARM development. The result is an increased safety during close interaction with patient and user (Haddadin et al., 2007). The reflected motor inertia is reduced by torque control up to a factor of 5 (Albu-Scha¨ffer et al., 2007), while the link inertia is low due to the lightweight design. Compact and slim design simplifies the integration of one or multiple ORTHOROBOTICS ARM robots in the crowded operating room. By adding specialized instruments and modifying the application workflows within the robot control, the ORTHOROBOTICS ARM robot can be adapted to many different surgical procedures. This versatility has been achieved by the design of the robotic arm itself and by the flexibility of the robot control architecture. To equalize the deficits of the lightweight approach in absolute accuracy, the ORTHOROBOTICS ARM robot control offers the possibility to integrate external position sensors (PSs) to close the position control loop. In almost every surgical procedure where high-absolute accuracy is needed, a procedure or/and an imaging systems for registration of the patient on pre-recorded diagnostic data are state-of-theart. These systems (e.g. navigation systems) can be used as external PSs for the ORTHOROBOTICS ARM robot control to achieve a highabsolute accuracy with the robotic system, whereas latency and low-sample rate of available medical tracking systems limit the dynamics of the task. In the understanding of the ORTHOROBOTICS ARM developers, the four basic parts of a surgical robotic system can be described as:

1 surgeon – medical know how, surveillance, flexibility, responsibility; 2 robot – exact relative positioning, endurance; 3 instruments – adaptation to a specific surgical task, dexterity; and 4 navigation – exact acquisition and planning of absolute positions.

Beside classic position control, the robot can be guided by the user through exerting forces by hand on the robot’s structure. This is implemented by using integrated torque sensors (TQs) in all joints and torque control methods. The control method is described in Albu-Scha¨ffer et al. (2007) as soft robotics. While haptic constraints in robotics are usually implemented on a position/velocity interface, the soft robotics concepts address robots with torque interface, where compliance is implemented at joint level, but also in Cartesian coordinates. With the different control modes the following robotic control architectures are possible:

• One or more ORTHOROBOTICS ARM robots are tele-manipulated by the surgeon from a distant control interface for example in endoscopic surgery.

• The robot is equipped with an instrument (e.g. drilling machine) and is guided by the surgeon manually, whereas the robot limits the free motion based on preplanned virtual spaces and directions in the form of haptic paths, targets, virtual walls (Ortmaier et al.,2006).

• The robot is equipped with an instrument (e.g. tip of a passive mirror arm for laser osteotomy) and performs a planned trajectory autonomously and is registered by additional sensors/procedures.

Two exemplary soft robotics applications have been implemented already on the first generation, the KINEMEDIC, which is in the commercialization phase at BrainLAB/Germany. Integrated into BrainLAB’s surgical navigation system, the KINEMEDIC prototype performs biopsies in cranial neurosurgery and drills holes for pedicle screws in spine stabilization. In both applications, the robot guides the surgeon to the point of interest (tumour or bore axis) by haptic means, whereas in the final step the robot simply holds the correct position and the surgeon completes the task manually. Instead of performing the robotic task completely autonomous, this interactive approach can enhance the acceptance of the robotic system as described in Jakopec et al. (2002). Additionally, different setups in the operating room can be combined with these control architectures. Owing to the lightweight approach, programmable gravity compensation and the corresponding dimensioning of the robot’s joints, the ORTHOROBOTICS ARM robot can be mounted hanging from a ceiling stand (Frumento et al., 2006), on a mobile cart or at the side rails of the operating table, as shown in Figure 1. The assignment of control modes, number of robots and the different setups is interchangeable.

2. Robot system design Beside the lightweight approach, one essential design criterion of the ORTHOROBOTICS ARM robot is the versatility of the system. It is designed to fit seamlessly into existing surgical procedures and clinical environments, but is also configurable and extendable to comply with rapidly changing development in medical treatment and safety. To achieve this challenging demand, the ORTHOROBOTICS ARM robot has been conceived by clearly distinguishing between platform and application according to the principles of platform-based design (Sangiovanni- Vincentelli et al., 2004). In this context, the application is designated to configure and parameterize the platform for a specific task. To achieve the demanded versatility of the system, it is necessary to cover as many system constraints as possible by the configuration, due to the faster optimization cycles in application design. Only absolute necessary, static system constraints which are inherent to the field of use (surgery) or unlikely to change have been assigned to the platform. All remaining constraints are classified as open and can be realized within the application. In order to avoid limitations in the number of valid applications, the ORTHOROBOTICS ARM platform design encapsulates all fundamental functions of the system into small, decoupled and tested blocks. The communication infrastructure of the ORTHOROBOTICS ARM platform grants low-level, direct and independent but yet convenient access to these blocks and is open for future extensions of the platform. Hence, application design can start directly on the hardware level and is only limited by the set of basic functions of the platform. The extent and impact of this design approach in the ORTHOROBOTICS ARM robot can be illustrated by some of the assignments of open and static constraints to application and platform, as shown in Table I. The ORTHOROBOTICS ARM platform enables motion of the TCP in six degrees of freedom (DoF), restrictions deriving from the surgical procedure can be considered as being part of the application. In endoscopic surgery, instruments are inserted into the patient’s body through small incisions or orifices, which limit the motion of the instruments to four DoF. Different platform-based solutions have been introduced to restrict the instrument motions according to this limitation either by specialized mechanisms (e.g. daVinci system) or by additional passive joints (e.g. Aesop, Zeus). In contrast, the ORTHOROBOTICS ARM robot offers the possibility to solve this problem within the application. One way is Cartesian position or impedance control about a virtual point of entry, which is assumed stationary or detected by sensors. A second approach is to control two joints of the robot to output zero torque. The last two entities in Table I partitions the safety of the system into basic safety functions of the platform (sensory and communication redundancy, safe stop function, etc.) and the combination of these functions by the application. This approach is necessary due to the different existing and oncoming safety standards in the field of medical robotics.

By methods of simulation (Konietschke et al., 2004), cadaver testing (Ortmaier et al., 2006) and interrogation of clinicians, sets of system requirements for a variety of challenging medical applications were identified. The range of applications includes endoscopic heart surgery, drilling for pedicle screws in spine stabilization, biopsy of tumours in neurosurgery and osteotomy with CO2 lasers in maxillofacial surgery. After partitioning these constraints either as static for the platform design or as open realized by the application, a set of design criteria have been formulated for the ORTHOROBOTICS ARM robot:

• slim, compact design to reduce the fear of close contact and to simplify the integration into the operating room; • lightweight to reduce the impact of collisions and to facilitate the setup in the operating room;

• redundant number of joints to enhance flexibility and to assist collision avoidance of the robot’s links with other objects;

• integrated sensors to measure all relevant physical properties of the system;

• different control modes where the user can always be in charge;

• motor, sensor and communication electronics integrated into the robot arm to reduce the internal cable harness and therewith the robot’s dimensions;

• pc-based external robot control for scalability reasons;

• re-configurable electronics and robot control for rapid application prototyping; and

• open communication concept to integrate external sensors and actuators.

Hardware design

3.1 Kinematic design

The issue of acceptance of new technologies is important for an improved learning curve of clinicians and technicians. Close interaction with technical systems demand understanding of the system, thus a central design issue of the ORTHOROBOTICS ARM robot is an inherent predictability of the system’s actions for the user.

3.2 Mechanical design

The integration of all hardware features established by the LWR (torque and PSs, safety brakes, integrated electronics, etc.) in a yet more slim design with dedicated groups of joints with intersecting axes is the most challenging aspect in the mechanical design of the ORTHOROBOTICS ARM robot. The design approach of the LWR, with a sequential chain of revolving single joints connected by shell structures (Hirzinger et al., 2002) turned out to be not feasible for the ORTHOROBOTICS ARM robot.

In respect of good accessibility to the integrated electronics and the upper arm and forearm being the only places to integrate the electronics, this approach is not scalable to the desired slim dimensions of the Orthorobotics Arm.

3.3 Electronic design

In contrast to the more physical-oriented view of mechanical designers, electronic designers consider robots as signaloriented devices. For electronic designers, robots are a composition of interfaces to the physical world, i.e. sensor and actuator modules (AMs), which are connected to computing resources by a heterogeneous communication infrastructure with strict performance constraints. Signal noise, cabling effort, and EMC influences are reduced radically by digitalizing signals near by the physical interface. Therefore, the electronic modules have to be tightly integrated with the mechanical structure. The design of sensor and actuator components is not isolated from the mechanical design, since the outline of the modules has to fit into the morphology of the robot without constraining it.

Moreover, adequate modularisation and the standardization of communication interfaces increase reusability and reduce design costs. Related functionality should be implemented as components using a common hardware platform. As a consequence, the art of sensor and actuator design for highly integrated robots is to find the best trade-off between high integration and efficient implementation, to meet the strict constraints, and determine a suitable modularization.

4. Infrastructure design: hardware integration and software drivers

Infrastructure of a robotic system enables the integration of the distributed electronic hardware, such as sensors and actuators, with control applications. Therefore, it has to provide convenient interfaces for application developers and constitute efficient but transparent glue between the distributed components of the robotic system. Thus, infrastructure involves the operation environment for sensors and actuators, communication protocols and software drivers.

Infrastructure design for the ORTHOROBOTICS ARM robot is determined by two opposite prerequisites: on the one hand, infrastructure should be an open platform that can be configured for versatile applications. This asks for a flexible and configurable architecture that is rich in functionality. On the other hand the electronic components are to be highly integrated with the mechanical structure. Thus, infrastructure needs to be lean and efficient to honor the sparse resources.

To balance these opposing requirements, a hierarchical approach is used: close to the mechanical structure, where available space is sparse, components are most dedicated. Higher layers successively add flexibility to the system leading to the use of general-purpose architectures on the top layer

4.1 The platform architecture

To balance the opposing requirements of flexibility and efficient implementation the ORTHOROBOTICS ARM platform is designed as a layered architecture, starting at dedicated highly integrated components growing towards general-purpose commercial of the shelf (COTS) platforms (Figure 5).

The applied communication protocols follow the same hierarchy as the modules: close to the hardware, lowbandwidth protocols featuring small implementation footprint are used, while the general-purpose platforms are connected by high-bandwidth standard communication protocols.

The modular layout on each level together with the aggregation of components on successive levels by the means of suitable communication creates the desired platform flexibility. The four layers sensor/AMs, joint modules, realtime hosts, and auxiliary hosts have proven to be a suitable hierarchy with sufficient flexibility at affordable design costs for robotic systems.

4.2 The system architecture

Following the strategy to implement as much functionality as possible on general-purpose architectures all control algorithms are implemented on the external controller host, which is a standard PC platform running the real-time OS QNX (QNX Software Systems, www.qnx.com). This approach allows the use of conventional development tools such as Mathwork’s Matlab Simulink and enables efficient algorithm and application development. Moreover, one directly benefits from the constantly rising computing power of COTS platforms.

5. Robot control

As mentioned before, the ORTHOROBOTICS ARM robot has two main control modes. Mode selection is accomplished in the application (Figure 9). The first one is the classical position control mode, in which the robot follows the commanded trajectory in Cartesian or joint coordinates as accurately as possible and is controlled to overcome external disturbances. This mode is required for exact positioning applications, such as laser cutting or bone drilling. The second operation mode is the compliant one, i.e. the so-called soft robotics approach, briefly described in the following section

5.1 Torque and impedance control

In this mode, the user can guide the robot to a desired position or on a desired trajectory by hand, in a highly intuitive manner (Albu-Scha¨ffer et al., 2007). This is done in the so-called “gravity compensation mode”, in which a torque controller on joint level reduces the effects of friction and provides the torques needed to sustain the own weight of the robot and the tool. In this free-floating mode, several virtual springs based on virtual potential fields are used.

6. Summary

The ORTHOROBOTICS ARM presented in this paper is a new versatile lightweight research robot tailored for surgical applications. The compact and slim design, realized by the design of differential couple joints and highly integrated electronics, allows close interaction of robot and surgeon directly at the operation table, where space is sparse. With its low weight of 10 kg, the accelerated masses are reduced significantly in comparison to an industrial robot, which enhances the safety in the unstructured environment of an operating room both for the user and the patient. Despite its low weight, a maximum payload of 30N and a workspace similar to those of the human arm has been achieved. By this the ORTHOROBOTICS ARM will be capable to accomplish a broad range of possible surgical applications. The integration of joint side position and TSs beside the common motor PSs, enables impedance and position control on joint and Cartesian level. Based on the ORTHOROBOTICS ARM, we are currently developing a prototype setup for endoscopic heart surgery consisting of three robot arms equipped with active endoscopic instruments and camera. This setup will be presented on the AUTOMATICA fair 2008 in Munich.

ORTHOROBOTICS® ARM Assembly

Figure 1: Orthorobotics Arm kinematics (left); overlay of extremal pitch joint configurations with plot of maximum reach (right)

Figure 2 Shoulder joints 1, 2, 3 (left), elbow joints 4, 5 (middle) and wrist joints 6, 7 (right) of the Orthorobotics Arm

Figure 3 Orthorobotics Arm equipped with passive linear guide and a drill machine (left); with endoscopic instrument in two extremal positions (right)

Figure 4 The hardware architecture features three layers of communication: BiSS, SpaceWire and Ethernet

Figure 5 Orthorobotics Arm joint module: an open platform that allows the easy adaptation to changing application and safety requirements

Figure 6 The layered Orthorobotics Arm system architecture: only the current control is implemented on the joint modules

Figure 7 From the signal path to the middleware view: the signal path mapped to the robot hardware yields signal-oriented components (i.e. the joint modules)

Figure 8 The controller structure for the DLR MIRO

Figure 9 The structure of the MIMO state feedback controller for the coupled joints

References

• Albers, J., Schmidt, T., Hassfeld, S., Heid, F. and Vahl, C. (2007), “Sternotomy and craniotomy by an autonomous robot”, Zeitschrift fu¨ r Herz-, Thorax- und Gefa¨ßchirurgie, Vol. 21.

• Albu-Scha¨ffer, A., Ott,Ch. and Hirzinger,G. (2007), “A unified passivity based control framework for position, torque and impedance control of flexible joint robots”, International Journal of Robotics Research, Vol. 26 No. 1, pp. 23-39. • Benveniste, A. and Berry, G. (2001), “The synchronous approach to reactive and real-time systems”, Proceedings of the IEEE, Vol. 79, pp. 1270-82.

• Butterfaß, J., Grebenstein, M., Liu, H. and Hirzinger, G. (2001), “-Hand II: next generation of a dextrous robot hand”, Proceedings of the 2001 IEEE International Conference on Robotics and Automation, Seoul, Korea.

• Craig, J.J. (1986), Introduction to Robotics: Mechanics and Control, AddisonWesley Longman Publishing Co., Boston, MA.

• ECSS (2003), ECSS E-50-12A SpaceWire – Links, Nodes, Routers and Networks, European Cooperation for Space Standardization, Noordwijk.

• Frumento, S., Michelini, R., Konietschke, R., Hagn, U., Ortmaier, T. and Hirzinger, G. (2006), “A co-robotic positioning device for carrying surgical end-effectors”, Proceedings of ESDA 2006, Torino.

• Green, P.S., Hill, J.W., Jensen, J.F. and Shah, A. (1995), “Telepresence surgery”, IEEE Engineering in Medicine and Biology Magazine, Vol. 14 No. 3.