Medical manipulators form a class of applications with typical characteristics and specifications. Among these medical manipulator types is a group of systems where X-ray beams are positioned with respect to an area of interest in or on the patient. The resulting image is used by the medical staff to perform a medical intervention on the patient. These interventions nowadays include treatments for diseases with e.g. a cardio-vascular, oncology or neurological nature. The trend is towards more complex procedures involving multi-vendor equipment. This leads to enhanced specifications on the medical manipulator with respect to payload, accuracy, speeds etc. Additionally, the medical intervention profession not only carries out standard procedures on non-standard equipment configurations. New procedures are researched for across the globe and adopted faster and faster. This requires that medical equipment adaptations are required at a faster pace too and that medical equipment diversity increases exponentially. In order to serve that market, the architecture of medical manipulators needs to be such that multiple configurations can be defined, built, tested and characterized quickly. A modular approach in control and advanced automation in commissioning is the only way this can be done in a financially healthy way.
In general, the need for sub-micron accuracy and high bandwidths is not felt in these application areas, contrary to e.g. CNC and semiconductor equipment. However, medical manipulators stand
amidst medical staff and are used in (surgical) procedures on humans. This imposes stringent demands on safety, reliability and constant performance over time than regularly found in industrial
applications. The aforementioned demands on system level directly translate to demands on motion control level. Therefore, a medical manipulator requires motion control that allows easy
generation of configurations and a safe and foremost robust control strategy.
Optimal tuning of the motion control loops depends on the attached mechanical load. Its mechanical properties such as mass, damping and friction determine the overall dynamics of the whole kinematic chain. Load parameters are seldom exactly known in practice. Moreover, its influence to the overall dynamics of the plant is difficult to predict without performing extensive numerical modeling. The value of extensive modeling is undisputed but the relevance for finding the optimal control settings is somewhat limited. Control parameter value determination is typically performed manually for a particular manipulator configuration leading to long commissioning times and suboptimal performance, especially because of the demand for robustness. Algorithms for load identification and self-commissioning of drives can solve this issue. Particularly, these kind of algorithms are needed to maintain performance if the mass distribution of the equipment changes in the field. Imagine e.g. a doctor adding specific equipment to the manipulator, needed for a specific treatment.
A single set of control parameters cannot cover the whole range of applicable loads. Although the demands in terms of speed and bandwidth may be limited, robustness is extremely important. This robustness has two faces: Robustness with respect to control performance and robustness over a longer period of time (months, years). Both are essential in the medical market. The robustness demand leads to systems being tuned conservatively, thus sub-optimally. Proven robustness is also what makes the use of a totally human independent auto-tuner in the field questionable. An intermediate solution is to apply multiple sets of controller settings that can be switched dynamically, with guaranteed controlled switch-over performance. Diversity in manipulator configuration leads to a frequent control loop tuning activity which is time consuming. As this process involves highly trained specialists, it becomes clear that this activity negatively influences time to market, evidence of robustness, flexibility in configuration management and error free machine commissioning.
A separate challenge relates to machine up-time. Medical equipment must be available 24/7 as emergency treatments occur constantly. A medical manipulator is, like any other machine with mechanical movement, subjected to wear. Therefore, maintenance is necessary. This scheduled down time requires careful hospital planning and is a particular unwelcome necessity in facilities that have only a single interventional system. For those hospitals, manipulator maintenance means rerouting of (returning) patients and redirecting emergency ambulances. Downtime as a consequence of machine failure and maintenance therefore directly affects available patient care and must be kept to a minimum. Regular maintenance is nowadays planned, often conservatively, as it is still preferable over machine failure during operation, as this can have direct lethal consequences. Machine uptime can thus benefit if maintenance is performed only when absolutely necessary. Machine performance monitoring and - conditioning are the inputs to achieve this.
A commercially available medical manipulator with at least 5 degrees of freedom will be used for the demonstration of advanced motion control concepts and machine monitoring as developed in terms of the I-MECH platform. The device has several parts that display a non-linear and non-time invariant behavior. The manipulator also involves a distributed control architecture and motion controllers capable of dynamically switching between multiple control settings.