MIniature Co-operative RObots advancing towards the Nano-range

A 10x10x0.5mm piezoelectric actuator, a motor, which has bimorphs performing a walking motion have been developed. By using multilayer technology low drive voltages, easily delivered by integrated circuits, are possible. The monolithic approach results in a thin robust structure without need for complex assembly. Depending on how the bimorphs are arranged rotational or linear motion with nanometre precision and mm/s speeds can be generated. First prototypes have been used for the MiCRoN robot.

Mobile microrobots:EPFL has developed several cm3 mobile micro-robots powered with piezo-actuators. They have a locomotion module with 3 degrees-of-freedom (x-y-Teta-z) and can be equipped with various tool modules, such as micro-gripper, AFM probe, micro-syringe, The robots are characterized by their high resolution (a few nanometers), compactness (cm3) and simplicity. They are composed of very few mechanical parts and can be produced at very low cost and are easily custom-tailored. These robots are also vacuum compatible, making it possible to use them in a scanning electron microscope. Potential Commercial Applications: - Micro-assembly (micro-optical components alignment, MEMS,); - Microscopy (light microscope, scanning probe microscopy, electron beam microscopy, focus ion beam, ); - Micro-manipulation (life sciences, micro-biosensors positioning, fertilization in vitro, ); - Instrumentation (Nanotribology, Material testing, IC prober, Nano-handling, ); These robots are being used in various micromanipulators and have been presented in several conferences and fairs. EPFL can provide consulting support to companies willing to integrate these micro-robots in their manipulation systems. On-board electronics are being developed to make these micro-robots fully autonomous.

A rapid prototyping technology for making multilayer piezoceramic devices has been developed. Advanced monolithic devices can be made in a short time and with only very small resources, which makes it possible for more companies or laboratories to fabricate such devices. The process is suitable for research and development or where only small series are produced, but the results can easily be transferred to a large-scale production process utilizing conventional tape-casting and lamination process.

A robust and at the same time flexible position control is a precondition to accomplish complex tasks. In order to get the robots follow the desired trajectory several issues needs to be addressed: - Different robots (Miniman as camera robot, different MiCRoN robots) have completely different movement behaviours - The dimensions and type of the drive command differ (e.g. different number of D.O.F.s) - Parameters are not constant over time and may lead to changing conditions - The movement is highly non-linear and tends to become chaotic at higher velocities using Piezo-driven robots. - The AEM - depending on the operation mode - needs a reliable underlying control system; two closed loops working at the same time have to be avoided. These reasons have motivated a two-staged approach for the position control. In this scheme, a driver that works in open loop only equilibrates non-linearities and balances the differences between different robots. At a higher level, a closed loop controller can base on a more or less linearised driver and is in charge of compensating the remaining deviation.

The Monolithic Piezoactuators (MPA) combines nanometre accuracy with centimetre motion range. Velocity up to several millimetres per second is possible for a thrust force up to a few Newtons. These performances are obtained at a significantly lower cost than with standard technologies and in a much smaller overall size. These actuators are extremely simple to manufacture and easily adaptable to different configurations and customer requirements. Only one electrical channel per degrees-of-freedom is necessary. MPA can be used for virtually any applications where relatively large workspaces must be combined with very high accuracy. Compared with existing systems, their main advantages are the high resolution, high stiffness, compactness and simplicity. MPA are easily custom-tailored. Innovative Aspects: - the actuator's configuration is obtained simply by the lay-out of the electrodes; - several degrees-of-freedom can be embedded in the same actuator. Main Advantages: - compactness; - high resolution over a large range; - vacuum compatible; - cost; - adapted to mass production. Potential Commercial Applications: - Micro-assembly (micro-optical components alignment, MEMS, Mask alignment, ); - Microscopy (light microscope, scanning probe microscopy, electron beam microscopy, focus ion beam, ); - Micro-manipulation (life sciences, micro-biosensors positioning, fertilization in vitro, ); - Instrumentation (Nanotribology, Material testing, IC prober, Nano-handling, ); - Mini-invasive surgery (micro-electrode feeder, micro-endoscope, ); - Micro-machining (EDM, maskless lithography, shadow mask, ); -

A micro-soldering demonstration was developed which would allow the MICRON robots to cooperate in soldering a 150 x 300 micron resistor to a custom-made simple micro-circuit. The basic steps of the process involve:(1) placement of the micro-circuit on a micro-hotplate, of size 750 x 750 microns (2) placement of solder pieces on contact pads by the robot, solder is with integrated flux (3) the resistor is placed on the contact pads (4) the micro-hotplate is activated, melting the solder, and bonding the component (5) the completed circuit can be removed from the hotplate. These steps are performed with the help of the robots, which place the various components of the process. A key factor of the process is the use of squares of solder of size 50 -120 microns which have flux contains within. This ensures that during the soldering process, the surface is contaminant free. The resistor used in the soldering process is the smallest SMT component available from a company, which manufactures these, and there is definitely the ability to place components of even smaller size.

Result description: In task 2.4 partner Fraunhofer-IBMT has developed a module for wireless powering of micro robots. Details of the results are disclosed in deliverable D2.7. The module for wireless powering consists of two sub-components. The first sub-component is located at the robot and is an integrated power unit. It consists of a miniaturized coil and an electronic circuit. The second sub-component is an external unit (power floor), which wirelessly transmits electrical energy to the integrated power unit by means of an external alternating magnetic field. The external magnetic field is generated by a special arrangement (patent pending) of inductive coils. The coils are fed by two alternating currents with fixed phase shift of pi/2. These currents are generated with a signal generator and two power amplifiers. The special coil arrangement assures that a significant magnetic field is only present within a small vicinity of the power floor. For receiving the inductive power, the robots must be located close to the power floor. The total area of the power floor is 24 x 24cm2. The available working area is about 20 x 22 cm2. The module for wireless powering is able to transmit 330mW power to the robot without heating up the power floor above 37°C. It fully fulfils the defined requirements. Dissemination and use potential: Fraunhofer-IBMT has filed a German patent on the power floor entitled "Vorrichtung und Verfahren zur induktiven Energieubertragung" (Translation: "Device and Method for inductive energy transfer") (Number: 103473DE, priority date: 8. June 2004). Additionally two papers covering the topics of MICRON-research have been presented during scientific conferences. A paper dealing with the power floor has been submitted to a scientific journal. Scientific publications: J. Brufau, M. Puig-Vidal,, J. Lopez-Sanchez, J. Samitier, N. Snis, U. Simu, S. Johansson, W. Driesen, J.-M. Breguet, J. Gao, T. Velten, J. Seyfried, R. Estana and H. Woern: "MICRON: Small Autonomous Robot for Cell Manipulation Applications". Proceedings of 2005 IEEE International Conference on Robotics and Automation. April 18-22, 2005, Barcelona, Spain Jianbo Gao: "Inductive Power Transmission for Untethered Micro-Robots", The 31st Annual Conference of the IEEE Industrial Electronics Society, Sheraton Capital Centre, Raleigh, North Carolina, USA, November 6 - 10, 2005, accepted as oral presentation Jianbo Gao: "Traveling Magnetic Field for Homogeneous Wireless Power Transmission", IEEE Transactions on Power Delivery, submitted on June 30, 2005 Key innovative features: Usually two coils are used for inductive powering transmission, the sending coil and the receiving coil. The straightforward solution for inductive power transmission is using two circular coils as the sending and receiving coils. This solution has some drawbacks because the sending coil in the power floor is required to be large (24 x 24cm2) and the robot coil must be about 1cm in diameter. In order to induct enough power in the robot coil, the sending coil must produce a strong enough magnetic field. That means there must be a strong current in the sending coil. Consequently, either the voltage between the two ends of the coil would be very high (> 10,000V) or the heat power of the coil is very high. This makes the straightforward solution not feasible. Therefore, an alternative approach for producing alternating magnetic field on the power floor was developed, which is called parallel coil method and which is described in deliverable D2.7. The innovative design and arrangement of the windings of the power floor allows running the power floor at a reasonable voltage. By transmitting the requested power of 330mW to the robot the temperature of the power floor does not raise above 37°C. In a plane parallel to the surface of the power floor the distribution of the magnetic field is homogeneous. A significant electromagnetic field is present in the close vicinity of the surface of the power floor only. Thus the developed power floor does not influence any (electronic) devices in its neighbourhood. It is planned to exploit the developed power floor. But up to now no company has been found who is interested in the power floor or at least in the new concept for inductive power transfer. During the next year partner FhG will intensify these efforts. The filed patent presents a good starting point for negotiations with companies. Current status and use of the result: At the moment the developed power floor is used to power the developed micro robots Expected benefit: It is envisaged to use the power floor for wireless powering of industrial micro robots if such robots are available in the future. It is also possible to use the power floor concept for wireless powering of medical implants.

The main purpose of the simulation unit is to verify the feasibility of the input task sequence to be autonomously and cooperatively executed by the robot agents. Simulation can detect two types of infeasible plans. First, infeasibility that might occur due to discrepancies between the task requirements and the hardware availability. Second, the infeasibility that might occur due to workspace and obstacle constraints. If the Task sequence of the Plan is validated by the Simulation module, then a set of goal configurations and tool commands are dispatched to the Autonomous Execution module for on-line autonomous sensor based motion planning and control. The main units of the Simulation module are the Parser, the Task Supervision unit, the Motion Coordination and the World Model unit. The software architecture of the simulation module was implemented using C++ object oriented programming.

In order to drive the different piezoelectric actuators mounted on MiCRoN robots, a 20V trapezoidal (for EPFL actuators) or saw-tooth signal (for DMS actuators) must be generated and applied between the actuator terminals if a deformation of this materials and therefore movement is desired. For this reason, a full-custom mixed-signal IC called PAA (Power Addressing and Amplification IC) has been designed using a 0.7¿Ym BCD (Bipolar, CMOS, DMOS) I2T technology from AMIS semiconductor that allows the inclusion on the same substrate of analogue high voltage transistors and digital CMOS gates. PAA IC contains 5 high voltage operational amplifiers (HVOA), which are internally set up in an inverting close loop configuration. The gain in each HVOA is fixed with a voltage divider based on integrated resistors. HVOAs have a power down system that it is enabled when the piezoelectric actuator controlled by the selected amplifier does not need to be driven. This fact allows the reduction of the global power consumption of PAA IC. The operation of PAA IC it is the following: a digital control unit decodes the control signals [cs1,cs0,b1,b0] that come from a main digital control unit IC (called MXS) based on a truth table that defines the different actuation modes that a MiCRoN robot can handle depending on the depending on the tool used by the robot. Since more than five piezoactuators have to be driven, two PAA ICs must be included on the electronics on board (Chip 0 and Chip 1) hence the digital control unit generates 10 internal signals [C9..C0] that enable (active high) or disable the 10 drivers included in Chip 0 and Chip1. An ID signal (Identification Chip) included in PAA IC allows selecting the internal enable/disable signals that will be applied to the five HVOA in each PAA IC. If ID=0, then the enable/disable driver signals are [C4..C0] whereas if ID=1 the enable/disable driver signals are [C9..C5]. This fact, allow us to implement a different control for the 5 HVOAs of Chip 0 or Chip 1 only setting up a different value for ID without any modification of the PAA IC design. When any driver on Chip 0 or Chip 1 is enabled (due to [cs1,cs0,b1,b0] from MXS) the digital control unit generates a RQ signal (Request signal) which means that the robot wants to make any operation with one of its tools. The RQ signals enables the awaking unit (but only in the PAA Chip where a HVOA has been enabled following the actuation mode truth table) that generates also an AWK signal (Awake signal) that goes to the start up unit. This unit starts a low slew-rate bias of the activated drivers in order to avoid great peak currents at the beginning that could damage the HVOAs. Once the selected amplifier/s are properly biased and ready to start their activity, the start up unit sends and ACK signal (Acknowledge signal) to MXS IC indicating that MXS IC can send low voltage (1V up to 2.4V) trapezoidal or saw-tooth signals that will be amplified (up to 20V) and applied to the selected piezoelectric actuators. The maximum working frequency of the amplified signals has been fixed to 1kHz for DMS material and 3kHz for EPFL.

The online multi robot controller is realized under the multirobot navigation functions framework. A special controller design that takes into account the micro robot motion principle is applied to increase the micro-robot motion accuracy. Multi-robot navigation functions are a special category of potential functions that have a unique minimum at the destination configuration. Their negated gradient vector field provides for a fast, feedback based closed form solution to the motion planning and the multi-robot motion control problem. This controller produces actuation signals in the favoured velocity region of each robot according to the vector field produced by the negated gradient of the multirobot navigation function. It features theoretically established properties of global convergence and collision avoidance. The user only needs to define the desired destination configuration of the multirobot system. The coordination and cooperation between the robots is dynamically produced by the navigation vector field depending on the robot proximity relations. Due to the feedback based character of the utilized control law, the methodology is very robust to sensor noise and model uncertainties.

The objective was to design, manufacture and test micro-grippers capable of being assembled to a micro-robot for potential use in a 3D assembly task. The key components of the gripper are: 1. The piezo-electric actuator, which has advantages of low power consumption, low weight and high force. 2. The gripper arms, which consist of stainless steel fingers machined by EDM (Electro Discharge Machining) These grippers have performed several types of manipulation task such as pick and place of 30 micron spheres, to picking of 35um dia. wires and 150micron wide resistors. They have been used in the micro-soldering assembly task, which was another deliverable of the MICRON project, and the grippers were used to place components prior to the soldering process, and then used for removal of the finished work-pieces. These grippers were also used in vision recognition systems to track the gripper during manipulation of micro-objects, and they have been marked to enable better tracking by machine vision. This work has been submitted to the ICRA 2006 conference.

There are several tasks for which the end effector of the manipulator must follow a specific motion trajectory with predetermined velocity. Those tasks include cell injection, surface scanning, griping, etc. The desired trajectory is described by a set of via-points and velocities in the end effector coordinates. The Trajectory Tracking controller is implemented using a P type control law with a velocity feed-forward term. The trajectories for the 4 degrees of freedom are first defined in the end-effector space according to polynomial time-laws. The end-effector velocities are transformed to joint velocities using inverse differential kinematics. These joint velocities provide the feedforward term, which is added to the scaled position error term and results to the trajectory tracking control signal. The sensing is performed by the corresponding camera robots.

Micromanipulation platform based on several mobile micro-robots: For research in the domain of micromanipulation often a very high flexibility is required in order to be able to perform a variety of experiments with the same manipulation set-up. A micro manipulation platform based on small mobile micro robots offers this flexibility: a combination of several relatively inexpensive cooperative robots with several degrees of freedom and a variety of integrated tools can be arranged in any configuration possible in order to perform a wide range of very specific micro manipulation tasks. The proposed test platform consists of an XY stage on top of which several cm3-sized mobile micro robots are moving with 4 degrees of freedom (X,Y,Teta_z,Teta_x). Two cameras are integrated on the set-up: one giving a global top view and one giving a local bottom view. The image of the global top view is used for position measurement of the mobile robots by optical tracking. The local camera gives a microscopic view of the objects to be manipulated and the tool tips of the robot, just like an inverted microscope. The working area of the micro robots consists of a steel plate with in the middle an aperture that is covered by a thin glass plate. The samples can be put on the glass plate through which they can be observed by the local camera from the bottom. Several micro robots have simultaneous access to the samples from the surrounding steel plate. As the working area is fixed on the XY stage, the whole scene - samples and robots - can be moved in X and Y direction, in order to move the field of view of the local camera to a certain zone of interest. The tool of the robot is fixed on a rotational actuator (Teta_x) that moves the tool along a circular path within a vertical plane in order to adjust the height of the tool. The local camera is fixed on a linear actuator in order to keep the tool tip of the robot always in the focal plane, by maintaining the length of the optical path constant. Application fields: - micro-biology; - nanotechnology; - MEMs; - micro/ nano-electronics; - micro-assembly; - education; -

The Atomic Force Microscopy allows imaging and interacting with biological samples in the scale of nanometres and nN. The designed and implemented AFM tool consists in two parts: sensor and actuator. The sensor is a piezo-resistive AFM commercial probe, which is connected to the actuator by means of a specially designed holder. The actuator is a multilayer piezoceramic element, which can be moved in three orthogonal directions. The actuator consists on four piezoelectric stackers (dimensions: 2mmx2mmx3,5mm) allowing the position control of the AFM tip with a resolution in the range of manometers. The holder allows the coupling among the piezo-actuator and the AFM sensor. This holder is suitable for the space requirements and, in addition, it contacts with the sensor mechanically, without the needle of soldering, allowing an easy interchange of diverse AFM tips. Finally, a cantilever forms the classical AFM sensor with a sharp tip probe at its end and the laser force detection system. In our case we only use a self-sensing cantilever, which has silicon piezo-resistance deposited over it. When the cantilever bends, the strain over the piezo-resitance changes the resistance value in linear way (this is true only for low variations). With this system we reduce drastically the dimensions and the power consumption comparing to the classical AFMs. The driver and instrumentation for this tool fits in the onboard electronics result, but it's especially important the possibility of externally adjust the instrumentation system in order to allow the AFM tip interchange. The results of the calibration of the AFM system, comparing it with a typical AFM experience (using a laser for measure the cantilever deflection) making a force curve over a crystal sample are: - The electronic noise causes a 10% of error, which could be filtered in the software system. - The non-linearity is lower than 1.2%, which implies the system works with the same accuracy than the laser measurement.

Individual microbial cells may differ from each other in their genetic, biochemical, physiological, or behavioural properties. Recent advances in analytical methods and technologies have enabled microbiologists to resolve these individual cellular differences at unprecedented levels of detail. Methods capable of single-cell resolution have provided fundamental insights into the inner workings of microbes and their interactions with each other, with higher organisms, or with the environment. The nanoscale AFM working robot should allow getting information on the effects of the nanomanipulation process in real time and on the scale of affected region on processes such as cell growth, cell proliferation, cell communication, etc. Microrobots incorporing AFM allow: - Not only topological measurements are available (with nanometric resolution); AFM is also capable of complementary techniques that provide information on other surface properties (stiffness, hardness, friction, elasticity, conductivity, etc.). - Rather than drying the sample (as done in the other microscopes), one can operate AFM in aqueous solution (which allows the live of the cells) and nanomanipulate cells in their physiological environments and study biological processes in real time. The microrobot has two modes of AFM: nanoindentation and scanning. Scanning is obtaining and image of the surface. Nanoindentation consists of obtaining a force curve in one point of the bacteria. Repeating this nanoindentation experiments all over the sample we could work in the calle "Jumping Mode" which is the best way to image samples in liquids. The microrobots could cooperate for doing simultaneous measurements in different part of the bacteria (i.e. for measuring the mechanical effort transmission through a membrane) or cooperate with a commercial AFM microscope. It opens a full range of experiences impossible of make with traditional AFM equipments.

The controller IC (MXS chip) receives instructions and transforms them into control signals to perform the requested action. Actions include the generation of waveforms with different shape and frequency, controlling of the tools with the possibility to use an on-chip PID controller and sending information of the sensors and the microrobot state. These tasks could be executed in a micro-controller. However, a solution based on an ASIC has been selected because is better in terms of area and power consumption. The MXS circuit has been implemented with a 0.35m technology from Austrian Microsystems. The core works at 3.3V and occupies 16mm2. It has 84 pins, but only 27 are used to connect the core with the other robot components. The chip is composed of full custom blocks; specific logic implemented with standard cells and customized IP cores (A/D and D/A converters and a power on reset (POR)). The IPs have been modified to adapt them to the chip requirements. The controller architecture is a Global Asynchronous Local Synchronous (GALS) architecture to deal with the power restrictions. To develop the GALS architecture, the different clocks are generated by an on-chip programmable clock generator module from a 40MHz clock produced by an external crystal quartz. For each operation mode, the controller generates 1 to 4 control signals O1, O2, O3 and O4, which are amplified by the high voltage drivers (not included in the controller IC). These signals are linear combinations of trapezoidal or saw tooth signals for the carrier and the rotor, whereas triangular for the AFM tip. They are coded in 10 bits to achieve nanometric resolutions, with a maximum resolution of 512 samples per period. The frequency range goes from 0.1Hz for the AFM operation modes to 2kHz for the maximum displacement speed. The number of control signals depends on the piezoelectric element (only one can be active at the same time): 4 for the AFM and the carrier and only 2 for the rotor. The signals can have a phase shift of 0?, 90?, 180? or 270? between them. Programming all these parameters (waveform wav, amplitude Np, sampling frequency Ts, phase and number of samples of the control signals Ns) is possible to execute straight and circular displacements in the carrier mode or to move the arm to position the AFM tip in the rotor mode. The signals are generated in a full period and stored in one of the four independent 512x16 SRAM included. For the calculation of the waveforms, they are decomposed into straight lines and each segment is calculated by an optimised Bresenham algorithm implemented in hardware. After generation of the waveform signals a 16-bit data-path mixes these signals as a function of the operation mode and outputs them through four 10b D/A converters. During mixing it is possible to apply a programmable gain (Gi) and offset (OFFi) to the signals. The gain Gi and offset OFFi are applied to compensate to the first order the piezoelectric fabrication mismatches for the carrier, rotor, nanoidentation and scanning modes. It also allows compensating the offsets of the high voltage drivers. For experiences requiring a closed loop control a digital PID has been also implemented. The main characteristics for the hardware implementation of the PID are the low power consumption and flexibility. Flexibility is required because several tools in the robot use the same PID hardware (AFM and micropipette). The final architecture of the PID is completely configurable by accessing at the programation registers. The data-path has been reduced to only one 16b adder and one 16b multiplier. To allow the communication between the robot and a host computer an IRC protocol has been included. It implements protocol wrappers and unwrappers of the physical IrDA protocol layer. Featured data rates range from 9.6kbit/s (SIR) over 1.152Mbit/s (MIR) up to 4Mbit/s (FIR) for flexibility in bandwidth and power consumption. Several IRC submodules can be switched off in order to minimize power input. Additionally, a low power protocol derived from SIR is implemented. CRC sums of up to 32 bits guarantee data integrity. The receiver module RX Unit can adapt to initial frequency deviations of 5% in SIR and MIR, and 2% in FIR mode. Frequency drift, e.g. due to changing working temperature, are compensated for, automatically.

Result description: In task 3.2 partner Fraunhofer-IBMT has developed a micro-machined syringe chip to be used as a cell injection tool. A detailed description of the chip can be found in deliverable D3-04. The microfluidic SyringeChip monolithically integrates a micro needle, a thermo pneumatic micro pump connected to this needle, and a sensor. The dimensions of the chip are 2.2 x 2.2 x 1mm3. The micro needle as well as the area around the micro needle is made of translucent materials (silicon dioxide, glass). This allows the observation of the cell injection procedure through the translucent chip (if requested). Filling of the chip is done by simply dipping the needle into the fluid to be injected. In contrast to commercially available injection systems the size of connecting tube and pump of the IBMT chip is adapted to the volume to be injected. Once filled, the chip can be used to perform several hundreds of injections. The injection volume can be adjusted and controlled very precisely. Power consumption of the SyringeChip is less than 2mW for up to 2pl injection liquid. The integrated actuator (micro pump) can be controlled by a PID-controller or even by a Pulse-Width-Modulated signal. The realised micro-needle has an outer tip diameter of 2µm and a length of 25µm. To be able to produce such tiny hollow needles a new plasma etching process has been developed. This process is able to produce needle shapes with tip diameters of even less than 2µm. The actuator is based on a low power consuming membrane-less thermopneumatic working principle, allowing to control very small fluid flows (» 0,2pl) through the needle. The used silicon processes are able to fabricate many SyringeChips in parallel. Dissemination and use potential: Three papers covering communication aspects of MICRON-research have been presented during scientific conferences. Additionally, the syringe chip has been presented at Biotechnica Exhibition, Hannover, Germany in 2003 and in 2005. In 2005 the syringe chip has also been presented at MEDICA Exhibition in Dusseldorf, Germany. Scientific publications: TAGLIARENI, F.; STADELBAUER, B.; TAHREEM, S.A.; VELTEN, T.: A microfluidic SyringeChip for Microinjection with integrated Actuator. In: Proc. of Int. Conf. Micro.tec. Munchen, Oktober 2003, S. 141-146 TAGLIARENI, F.; VELTEN, T.: Microfabrication of a microfluidic SyringeChip with integrated Actuator. In: Proc. of IEEE Int. Conf. Mechatronics & Robotics. Aachen, September 2004, S. 173-177 TAGLIARENI, F.; NIERLICH, M.; STEINMETZ, O.; VELTEN, T.; BRUFAU, J.; LOPEZ-SANCHEZ, J.; PUIG-VIDAL, M.; SAMITIER, J.: Manipulating biological cells with a micro-robot cluster. In: Proc. of the Int. IEEE Conference on Intelligent Robots and Systems (IROS). Edmonton, Alberta, Canada, 2005, S. 426-431. Key innovative features: They key innovation is to integrate micro needle, micro pump and all tubings on one tiny chip. This allows making the injection equipment very cheap. Additionally, the injection of tiny amounts of liquid becomes more reliable and reproducible because long and flexible tubings are avoided. The integrated sensor additionally adds to the reproducibility of the injection volume. This is a big step towards automated cell injections. Current status and use of the result: At the moment the developed syringe chip is used as a tool for micro robots. As an alternative to the use of the micron robots the syringe chip can be used with commercially available micromanipulators like the MM3a from Kleindiek Nanotechnik, Germany. It has been shown that the developed syringe chip together with the MM3a manipulator is able to perform cell injections. Diacetylfluorescein has been successfully injected into L-929 and HL-60 cells. Expected benefit: The developed microinjection chip has the potential to become a cheap alternative to cell injection equipment used nowadays. It is cheap and allows injecting tiny amounts of liquid in an accurate and reproducible manner. This is a big step towards automated cell injections.

Using the camera system developed during the MiCRoN project, a vision solution for the localisation and tracking of micro-objects in the field of view of the camera has been developed and implemented. The technique combines a geometric hashing algorithm for object recognition with up to 4 degrees of freedom with a fast tracking algorithm based on the bounded Hough transform. The combined algorithm is able to localise the position (x, y) and the orientation of an object (or part of it) on the image plane as well its displacement from the focal plane when this is confined within a limited range which depends on the optics of the camera. In order to achieve this, an object model comprising a stack of images has to be acquired offline and pre-processed prior to the execution of the recognition/tracking algorithm. The algorithms are built using a machine vision software library (Mimas toolkit) developed at the MMVL under the LGPL licence. This software toolkit allows the fast prototyping and testing of new machine vision algorithms in a wide range of applications, both in the industrial and the biological areas. These results have the potential to open up new markets and new opportunities throughout Europe in the application of machine vision to sector such as microsystems, microassembly and biological manipulation.

The user first decomposes the experimental procedure into an appropriate sequence of tasks. This task sequence is then input to the system through a Graphical User Interface (G.U.I), using a task description language specifically developed for the MiCRoN platform (Micron Task Language -MiTL). A lexical parser parses the input task description and retrieves relevant task data from a task bank to create a sequence of motion and tool commands. The commands are fed to a simulation stage for feasibility verification. Successful task sequences are routed for execution by the system controller. Motion data are wirelessly provided to the micro-robots and a sensor based fusion scheme is then applied to close the control loop. The main s/w modules of the architecture are: The Parser class: This class parses the input text. Generates appropriate data structures that encapsulate the goals and info related to the participating robots. TaskSupervision class: Implements the Task Supervision unit. Coordinator class: Is the class responsible for interfacing the navigation algorithm to the Supervision unit. TrajectoryTracking class: Is responsible for the trajectory-tracking algorithm. InputData class: Is the class responsible for storing goal configuration, tool commands and controller parameter data structures. SetOfExeptions class: Is the class that includes all the possible exceptions that can be generated by the Simulation Unit. RobotModel class: Incorporates information about the robot model like maximum velocities per DOF and kinematic models for each robot to actuate them. Several other subsidiary classes fill-in the software architecture.

The main challenge of the microrobot implementation is the size minimization with powering and communication autonomy. The electronic module is designed to perform it and the following predefined robot capabilities: read/write from/to PC the desired microrobot movement and manipulation with the robot tool, drive 10 different piezoelectric actuators with capacitance values of 10nF to 35nF with trapezoidal and sawtooth voltages waveforms up to 2.5 kHz, provide a force control system to work with some tools (specifically the AFM tip and the syringe, which need closed loop control). In order to accomplish these requirements, 4 different modules are designed and implemented: Power source generation module (PSG), Input sensing for control system (ISC), Mixed signal IC module (MXS) and Power addressing and amplification IC module (PAA). All these modules are full custom designed and placed in a 4 printed circuit boards of 12mmx12mm. The microrobot electronics needs three different constant voltages for work: 3,3V for the digital circuitry and the instrumentation system, 20V for the power circuitry actuation (driving power to the piezoelectric actuators) and 5V for the power circuitry control signals. Because of the powering microrobot system (battery or power floor), the input robot voltage could be in the range of 2V-4V. The full power source generation module is implemented in two printed circuit boards (one layer each one) with a size of 6mmx12mm (the first one) and 12mmx12mm (the second one). The PSG3.3 integrates a low power DC/DC Buck/Boost converter, which supplies 3,3V from an input voltage in the range 1.6V-5.5V. The nominal performances of the circuit are: output current 115mA; high efficiency (up to 80%); low quiescent current (50µA); low output voltage ripple (3.2% in the worst case). These nominal values comply with the microrobot requirements. It also complies with the dimensions requirements due to the capacity of the DC/DC for work without an external inductor. The PSG20 is also a low power DC/DC converter (Boost). The converter output voltage is fixed to 20V and the nominal performances of the circuit are: high efficiency (up to 77%); low quiescent current (28 µA); low output voltage ripple (2.7%). This PCB also regulates 5V for driving the control signals in the analogic ICs. Due to the dimensions requirements and the low power consumption for this control signals (about 1mW), a zener diode solution is implemented. The CPD module integrates a full custom mixed-signal IC (the robot ¿brain¿), two full custom power amplifiers ICs (for drive the piezoelectric actuators) and the system's clock. The ICs are soldered using wire-bonding interconnections. The full PCB is 12mmx12mm (one layer). The mixed-signal IC is designed using the AMS (Austria MicroSystem) 0.35µm microelectronic technology from Europractice service. This IC is the responsible of interfacing with the IR protocol, generating the appropriate signals (trapezoidal, sawtooth and triangular) for the microrobot movement and controlling the tool-closed loop by an externally programmable digital PID. The power addressing and amplification module is full custom designed using the IIT microelectronic technology from Europractice service. Using two ICs, the system is able to amplify the signals, which come from the mixed-signal IC, obtaining 10 independent power outputs, which are able to drive the microrobot piezoelectric actuators. The ISC Module is implemented in the two layers of a 12mmx12mm printed circuit board. Due to the kind of tools, which the robot has to sense (an AFM head, a micro-syringe and a micro-gripper), the instrumentation is designed for work with resistive. A Wheatstone's bridge solution has been implemented for signal conditioning, and an instrumentation amplifier is used to obtain the desired signal range to work with the software system range. This measure is fully differential, allowing work in noisy environments. In the experimental results we see about 10% of noise error, but lower than 1.2% when filtered in the host PC with a first order low-pass filter. Taking into account the possibility of interchanging tools to the microrobot and resistance's tolerance value, two multi-turn variable resistances are used to adjust accurately the Wheatstone's bridge (it could be adjusted externally, allowing the tip exchange). The instrumentation amplifier used fits with the system low noise (35nV/√Hz), low offset (250µV in the worst case), high gain (100dB) and small dimensions requirements. All this electronics modules have been interconnected using a 3D structure which is used also as a mechanical holding and make possible to minimize robot's dimension.

EPFL proposes a methodology to design micro-actuators. A family of piezo-actuators, especially well adapted for mobile micro-robotics, has been developed. These actuators feature very high resolution (in the nanometre range) with long motion range (several millimetres or even more). Velocity up to several mm/s is possible. These micro-actuators can combine several degrees-of-freedom. They are composed of very few mechanical parts and can be produced at very low cost and are easily custom-tailored. The proposed technology is compatible with mass production. These actuators are being used in various micromanipulators and have been presented in several conferences and fairs. EPFL can provide consulting support to companies willing to integrate these actuators to their manipulation systems.

During the course of the MiCRoN project, a compact microscope unit integrated on a mobile robotic platform has been designed and realised. The unit comprises the following components: (1) A carrier (MINIMAN-IV robot) with 3 DOF (x, y, and rotation around the vertical axis). (2) A linear (vertical) drive on the robot chassis to which the camera housing is attached. The resolution of this drive is about 0.1mm. (3) A camera housing containing the CCD chip and the miniature camera board for PAL signal generation. (4) A miniature lens that is able to provide a magnification of 5x. (5) A miniature stepper motor that is able to independently vary the position of the camera lens with respect to the imaging chip with a resolution of less than 1 um. The travel range of the lens support is 6 mm. This is accomplished by an aluminium bracket that translates the movement of the motor shaft into a linear (vertical) motion of the lens support. The combination of the two motors allows the camera to focus on an object present in its field of view (about 1mm2) placed at any height (within the range allowed by the mechanics of the MINIMAN-IV robot and its Z-drive). The depth of field of the camera optics is approximately 200-300um. The working distance between the front of the lens and the object is about 4-5mm. Although this imaging device cannot replace large and powerful optical microscopes used today in a large variety of application domains, it represents an alternative and cheap solution that can be effectively utilised in many applications where physical constraints in the work area and the requirement for a mobile platform make the use of this small unit preferable to a traditional optical microscope. In the near future, we intend to further develop our camera prototype using more recent miniaturised optical and imaging components as well as including new components such as an integrated LED light source for autonomous operation. At the same time, we will be continuing the development of the vision software to address current limitations as well as to make it portable to a variety of applications. The successful integration of these technologies and the realisation of such a flexible and portable imaging system will open up the way to possible commercial exploitation, especially if industrial partners with a strong interest in this research sector can be found. We are currently focussing in this direction and are confident that we will soon be able to attract industrial collaboration into this project.

Result description: In task 2.5 partner Fraunhofer-IBMT has developed a miniaturized wireless communication module for bi-directional communication with micro robots. The module for wireless communication consists of two sub-components: the micro robot transceiver and the external communication transceiver. The communication system between individual robots and the host was developed as a single-master multi-slave topology. Robots act as slaves to the host (master). For maximum mobility, wireless communication with infrared (IR) transmissions is used. The implemented protocol is modelled on the physical IrDA layer in order to achieve a robust data link. Featured data rates range from 9.6kbit/s (SIR) over 1.152Mbit/s (MIR) up to 4Mbit/s (FIR) for flexibility in bandwidth and power consumption. Additionally, a low power mode derived from the SIR protocol is implemented. CRC sums of 16bits for SIR and MIR, and 32bits for FIR guarantee data integrity. The receiver circuits can adapt to frequency deviations of up to 5% in all IR modes. The infrared communication implies that every data packet is received by all robots of the cluster. To address only one single robot, unique identification numbers (ID) are (dynamically) assigned to the robots. Data packets containing an ID are interpreted by the corresponding robot only. The micro robot transceiver is located at the robot. This on-board communication sub-system includes an ASIC module integrated with the ASIC chip of the robot electronics and a commercial IR-transceiver. The developed on-board communication sub-system fulfils all requirements. It has especially been designed for extremely low power consumption and for low size. The external communication transceiver can exchange IR-signals with the robots like described above. Additionally, the external communication transceiver has a wired USB2.0 link to a computer. A software driver has been complied under Linux system to provide a software interface for the upper level user interface. The full functionality of the communication has been tested using emulated robot electronics. The performed tests proofed a successful communication with the robot prototype on the power floor. Dissemination and use potential: Four papers covering communication aspects of MICRON-research have been presented during scientific conferences: NIERLICH, M., STEINMETZ, O.: "Manipulating Biological Cells with a Micro-Robot Cluster". IROS 2005 NIERLICH, M., STEINMETZ, O.: "A Monolithic Control Circuit for a 1cm3 Microrobot for Biological Experiments". Asian Solid-State Circuits Conference, IEEE A-SSCC, 2005 NIERLICH, M., STEINMETZ, O.: "An integrated Controller for a Flexible and Wireless Atomic Force Microscopy". Proc. SPIE 2005 - Microtechnologies for the New Millennium 2005 CASANOVA, R., LACORT, DIEGEUEZ, J., ARBAT, A., PUIG,M., SAMITIER, J.,NIERLICH, M., STEINMETZ, O., SCHOLZ, O.: "A Specific Integrated Controller for Nanomicroscopy and Cellular Manipulation". Oral presentation at Annual Conference of the IEEE International Society of Circuits and Sensors (ISCAS), Kobe (Japan), 23.-26.05.2005 Key innovative features: Commercially available hardware components have been used for the transceiver module. The key innovation is the development of an ASIC which contains the digital part of the transceiver module electronics. The transceiver module has been designed to be low power consuming and to require low space. Thus, it is well suited for applications with autonomous micro robots. Additionally, an USB2.0 interface to a computer has been implemented for the developed external communication transceiver. This allows a simple use of a PCs standard USB port to exchange data between PC and external transceiver unit at high speed. It is anticipated that the developed communication system can also been used for medical applications. Because of its small size and low power consumption (due to the dedicated ASIC) it seems to be well suited for the wireless communication between medical implants and an external control unit. The human skin is well known to be partly transparent for infrared light. A transcutaneous infrared communication is possible if the medical implant is not implanted too deep inside the tissue. Current status and use of the result: At the moment the developed wireless communication module is used for communication between micro robots and an external control unit (master slave configuration). Expected benefit: It is envisaged to use the wireless communication module to wirelessly transmit data between a control unit and industrial micro robots if such robots are available in the future. Maybe it is also possible to use this system for wireless communication with active medical implants.

For all microrobot handling, a so-called "Global Positioning system" must give the position of the robots to the robot control. At the University of Karlsruhe a position sensing system called "MPS" is currently being developed. With a resolution of 1µm and a turn-around-time of 0.2s this system is very exact and fast. Working touchless, its measurement principle works two-staged: the first approximation step is using the photogrammetrical principle, this results based on interferometrical effect delivers the mentioned resolution of 1µm. The freely movable robot is equipped with three marks working both as a photogrammetrical and as a Moire-based mark by having three or four coincidental cosinusodial shaped full circles printed upon it. A high-resolution CCD-camera is watching the whole scene permanently. Because of the Moire-effect, there are two grids needed to generate the Moire-fringes, which have to be analysed by the connected computer with the developed software. The first grid is the grid on the robot marks. The second grid, which is fixed to the working plate, is virtually created inside the computer (this grid is also circular and cosinusodial shaped). Merging the CCD-picture (after some image processing) with the virtual grid, the Moire-effect will appear. Using some more image processing, the intersection points of the Moire-Fringes are extracted. With a over-specified linear equation system, the centre point of the mark can be calculated then. Creating and solving the error function of this equation system, the resulting error is about 0.1µm within the resolution of 1µm.

Leistungen

Diese Seite teilen Diese Seite in sozialen Netzwerken teilen

Herunterladen Den Inhalt der Seite herunterladen