Microrobotics

Computer Scientist Sergej Fatikow on the variety of microrobots, possibilities for their application, and current research challenges

faq | June 13, 2018

First of all, we have to define „micro/nano robotics“. Is it about the robot’s size? Is it about the precision the robot can operate with? Or is it about the size of objects the robot can manipulate or handle in some way? Different research communities have different views that depend on the application – the ability of the robot, which is essential for the successful implementation of a particular task. Generally, “microrobotics” deal with small-scale robots, the size of which varies from micrometers to centimeters.

The basic classification of microrobotics focuses on the essential features of the microrobot, which are required for the aspired performance. On the one hand, we design and apply robots capable of manipulating or positioning with precision at the micro/nanoscale – high-precision microrobots. You deal with such robots if you are interested in handling a nanowire with a thickness of about 1000 times smaller than the thickness of a human hair or you want to touch an electrical path to do electrical characterization on a microchip, and the width of this path is just 10nm, or you try to implement injection or extraction into/from a biological cell. Moreover, you want to perform these and similar tasks in an automatic way and without tedious manual control, exactly as regular industrial robots operate in a factory. These examples show the idea behind high-precision microrobotics.

On the other hand, various research communities address applications where long-range mobility is the crucial feature of a robot. The “long-range” is to be seen in relation to the robot size, of course. For a micrometer-scale robot, the operation distance of several cm is a long-range distance. Those are mobile microrobots. For example, many labs look into the development of very small drones or very small crawling robots, or very small fish-like robots for monitoring in different environments. Medical microrobotics is another large research area dealing with robots operating inside the human body. Microrobots can use the natural pathways in the human body, like blood vessels or the gastrointestinal tract, for diagnosis and treatment to deliver drugs to a particular area of the body. The latter is part of so-called targeted medicine. These and similar applications are the focus of mobile microrobotics.

History of Micro- and Nanorobotics

Many communities refer to the famous speech by Feynman, “There is plenty of room at the bottom,” in 1959 as the birth of microsystem technology (MST) that later triggered the development of nanotechnology and bio-nanotechnology. The first ideas of microrobotics appeared with the emergence of an active investigation of MST. The rapid development of microfabrication technologies made it possible to provide components for microrobots at a sensible cost. On the other hand, very soon, it became clear that the development of MST and nanotechnology is hardly possible without high-precision microrobots enabling nanohandling.

Robotics expert Mel Siegel on the features that a machine must have to be called a robot, previous studies, and the future of the field
The need for targeted medicine was another favorable factor for microrobotics. The idea of microrobots autonomously operating inside the human body first appeared in science fiction movies like Fantastic Voyage (1966). With the rapid development of MST and nanotechnology in the 1980s-1990s, those ideas started to seem less fictitious. The progress in scanning electron microscopy (SEM) and in scanning probe microscopy (SPM) in the 1980s also had a large triggering effect on the development of microrobotics.

All this inevitably led to the consolidation of the new science of microrobotics and to forming related professional communities, which happened at the beginning of the 1990s. It means that microrobotics, as an independent subject of scientific investigation, has existed for 25-30 years now and gradually gained maturity. The feasibility of medical in-vivo microrobotics was demonstrated by experiments on animals. High-precision microrobots successfully implemented sophisticated and fully automated nanohandling sequences on nanowires, graphene flakes, nano-spheres, and other 1D-, 2D- and 3D objects at the nanoscale. Many labs deliver promising results on intelligent microrobot swarms, large amounts of mobile microrobots operating in coordinated ways to perform a monitoring task. Microrobotics is the best way to become a prominent member of the large robotic family that includes industrial robots, service robots, medical and rehabilitation robots, cleaning robots, humanoid robots, and others.

Structure

We are following the fundamental idea of robotics to mimic the functions of a human or of an animal. There are three major functions: to collect data from the environment (sensing), to process this data, to extract valuable information and make a decision (control), and to implement the decision and adjust the robot’s behavior (actuation). Obviously, the robots need the energy to implement these functions (energy supply).

The ultimate dream of any robotic researcher is to achieve fully autonomous behavior. It means that all four functional components are “on-board”, and the robot can operate without any external control. Humanoid robots or autonomous vehicles are good examples reflecting the current R&D activities. Microrobots are, however, limited in size, so it is not possible to integrate all the components. That is why in most application fields robots have to rely on external sensors.

Design

The design – and the corresponding fabrication method –depends on the size of robot components, which, in turn, depends on the application. Having in mind the large diversity of applications, it is not possible to give a simple recommendation on “how to design a microrobot”. A sensible approach would always be to start from a set of performance specifications. It will lead to a set of design constraints: robot size, movement range, force range, necessary sensor data, compatibility with environmental conditions, and so on. Then, one has to think about suitable actuation and sensor approaches, as well as about materials for the robot implementation. After we have clarified all this, we may start with design, having in mind the existing fabrication options.

Mars Science Laboratory Project Scientist Ashwin Vasavada on finding signs of a habitable environement, the Radioisotope Thermoelectric Generator, and Mars 2020
Important design feature, especially for high-precision microrobots, is the use of flexible joints fabricated by material tapering. Regular joints have non-neglectable mechanical play resulting in positioning uncertainty that can be comparable with the desired positioning precision. In such cases it is not possible to control the robot’s position, and the use of flexible joints is the only solution. The intrinsic re-store force based on mechanical properties (stiffness) of the material is the other essential feature of a flexible joint. The most frequently applied driving methods in high-precision microrobotics, the slip-stick drive, the inchworm drive, and the ultrasonic drive, are frequently implemented by piezoelectric actuators in combination with flexible joints. Many mobile microrobots mimicking insects, birds, fishes, and other animals (biomimetic robotics) are also frequently designed using flexible joints.

We also have to think about the sensor feedback while designing a microrobot. For example, if vision feedback from a microscope is necessary for the robot control, then the robot’s component that is being tracked (for a high-precision robot this is normally the end-effector) has to be always in the view field of the microscope. The robot has to be designed accordingly in order to ensure permanent visibility of the relevant component. This requirement is especially important for nanohandling automation inside an SEM as the field of view is extremely small, just a few mm2.

Materials

Material choice fully depends on the application and the robot size. Important criteria can be desired force, movement range, costs of fabrication, or compatibility with the application environment. High-precision microrobots need actuators featuring repeatable and controllable changes in position with extremely high resolution, up to the sub-nanometer scale. From this point of view, piezoelectric materials are usually the first choice. They are robust, powerful, precise and widely available on the market in different shapes and sizes. Silicon is the first choice if you need to apply microfabrication techniques due to the size requirements of the robot components. Polysilicon electrostatic actuators designed in comb-like shape and fabricated by surface micromachining are widely used if high forces are not required. There are many electrostatically driven silicon grippers for high-precision microrobots on the market. Shape memory alloys are another option, especially for applications that do not require a very fast response time from the actuator. Various polymers can also be used to implement actuators for microrobots.

If microrobots have to operate under non-ambient conditions, like in the vacuum chamber of an SEM, you have to take into account the environmental limitations. You need to protect the SEM chamber from contamination, and you need to prevent damage caused by the electron beam, and vice versa, negative effect on electron beam behavior. Material choice becomes especially tricky when microrobots are to operate inside the human body. Biological compatibility and non-toxicity are the obvious limitations, on top of all others. Unfortunately, a “perfect” material for microrobotics does not exist.

Control

The ultimate goal of any robotic technology is the highest possible level of automation. Open-loop control without sensor feedback is never possible under real conditions because the mathematical model of the control path would never accurately reflect reality. This is especially true for micro- and nanoscales as physical effects change dramatically with scale. There is still a lack of understanding of scaling laws, so object behavior cannot be predicted exactly. Therefore, the only way to control a microrobot is to close the loop and provide feedback on the current state of the control path using sensors of different nature. The current state is compared with the desired state. If a deviation is detected, then the controller sends instructions to the robot actuators in order to eliminate the detected deviation to make the robot follow the desired trajectory.

Philosopher David Edmonds on relationship between moral philosophy and robotics, artificial intelligence and ethics that we should build into machines
Vision feedback is the most widely used option in microrobotics. The technical implementation of the visual feedback differs depending on the required sensor performance. For handling microscale objects, the use of an optical microscope is usually sufficient. To handle nanoscale objects, high-precision microrobots require more powerful vision feedback. There are two options – SEM and SPM. The latter technology is slow and not suitable for robotic operation at high speed. This makes the SEM the only feasible technology to implement fast robotic nanohandling. In my lab, for example, we implemented automated assembly of so-called super-tips of an atomic force microscope (AFM). One option to improve AFM performance is to increase the tip sharpness by attaching a nanowire to the tip. Our high-precision microrobots performed this assembly automatically, using carbon nanotubes with a thickness of about 150nm.

Another example from my lab is the automated pick-and-place manufacturing of dedicated three-dimensional nanophotonic structures made of 400nm small spheres. The bottleneck on the way to high-throughput automation is the imaging speed, which is the common limitation of any scanning technology in contrast to optical imaging. Increasing the imaging speed of a scanning microscope (SEM or SPM) in order to provide fast feedback for automated robotic nanohandling is currently a serious R&D challenge.

Even more challenging is to receive fast vision feedback from mobile microrobots operating inside the human body. The most feasible options at the current state of research are the use of MRI-based tracking or electromagnetic tracking of magnetic parts of the robot. These approaches are, however, just at the beginning of the active investigation.

Energy Supply

Generally, energy supply is implemented differently depending on the robot’s size and performance specs. Larger microrobots allow for onboard actuation using piezoelectric materials, shape memory alloys, polymers, etc. Such robots are either wired or may use a battery on board. There are many interesting options for energy supply for so-called microswimmers, microrobots operating in a liquid environment. Contrary to onboard actuation, microswimmers can be driven by remotely generated forces. The most well-known approach is the use of external magnetic coils to power magnetic microrobots.

Other options include different methods of self-propulsion and methods utilizing the energy of biomaterials like cells or bacteria. The main challenge here is the predictability of the robot’s behavior. We still do not fully understand what is going on at extremely small scales. It is like a recipe including many ingredients: the robot’s shape, the robot’s size, the robot’s material, viscosity of the medium, temperature, and many others. You need to find out what is the perfect combination to ensure the robot’s controllability. In this domain, we need much more effort towards basic research, whereas high-precision microrobots with onboard actuation are about to repeat the success story of fully automated industrial robots within the next several years.

Application Field

In the opening ceremony at the first international conference on Manipulation, Automation, and Robotics at Small Scales (MARSS) in Paris two years ago, a representative from the industry, the CEO of a Swiss microrobotics company, delivered a greeting speech. He shared with the audience his memories of our discussions more than twenty years ago in Boston, MA, where a small bunch of scientists started having the very first annual workshop on microrobotics. One of the “crazy” ideas that we raised in Boston was the development of nanohandling robots operating inside the vacuum chamber of an SEM and using SEM images as real-time feedback for nanohandling automation. It was long before “nanotechnology” became a buzzword across various communities, so the impact of this approach was not obvious. Currently, there are numerous companies worldwide that have a successful business in high-precision microrobotics for nanohandling.

Physicist Seth Lloyd on quantum mechanics, classical vs quantum computers and the “universal digital computation”
The applications include assembly and manufacturing at small scales, manipulation of nanoscale objects, mechanical and electrical characterization at the nanoscale, handling and characterization of biological cells, including automatic injection and extraction tasks (so-called cell surgery), robot-based structuring of nanomaterials, measurements requiring positioning with nanoscale precision, and many others. Generally, we are talking about all kinds of operations at small scales where high-precision robots are involved or have to be applied because there are no alternative approaches.

The professional community primarily works on cargo delivery applications, including drug delivery and stem cell delivery to a particular region of the body. Another promising application is the monitoring of chemical and physical parameters inside the body. The researchers address not only obvious pathways inside the human body, like the gastrointestinal tract or blood circulatory system but also the central nervous system and eye.

One more prominent example is microrobotics for targeted medicine, including smart endoscopic pills, which can go all the way through your gastrointestinal tract to receive real-time feedback from inside the body, or externally operated magnetic microrobots propagating through blood vessels and deliver drugs exactly to the spot of interest. Even bacteria can be “trained” to operate inside the human body in a predictable way as a mobile microrobot. Commercialization of such technologies is more difficult than in the industrial domain and will take more time. However, the need is obvious, the interest from medical experts is enormous, and the ideas seem to be feasible from a technical point of view.

Challenges and Perspectives

The small scales are dominated by forces and effects that are very different from those that we experience at the regular scale. Gravity and inertial forces are of less importance or even fully negligible in some applications. On the other side, surface forces or friction forces have a dramatic influence on microrobot behavior. The problem is that we still do not fully understand the mechanism of this influence, so more basic research into these phenomena is indispensable. The general idea is to achieve the predictable behavior of a microrobot in order to be able to control the robot’s operation.

There are also application-specific challenges. For example, the next significant R&D step in industrial high-precision microrobotics will be the development of fully automated microrobot cells for high-throughput operation, using the feedback from an SEM. The bottleneck and the biggest challenge is here the speed of SEM imaging, including fast 3D localization. A serious challenge in targeted medicine utilizing mobile microrobots is to achieve predictable behavior of the robots, or controllability, including the control of swarm behavior. Biocompatibility and non-toxicity over a longer period are other challenges in this domain. Further, untethered microrobots for monitoring applications, both civilian and military, need a reliable, long-lasting power supply. Such applications also entail serious design challenges, depending on the deployment environment: flying, crawling, or swimming microrobots. Communication in microrobot swarms is another big issue in this application domain.

Making predictions about the future is a daunting task. I am very optimistic about the future of microrobotics due to the dramatic rise in the number of professionals dealing with this new science, which results in rapid development in many application domains. Microrobotics will follow the way of robotics on a regular scale. We should not forget that regular robotics has a history of active development, starting in the 1950-the 1960s. As a result, robotics became a highly developed technology in industrial manufacturing but has not yet reached its maturity in service robotics, medical robotics, and many other areas.

In the next several years, we will see serious improvement in the performance of microrobots and a rising number of start-ups that will use the potential of microrobotics to boost innovations. In some application domains, we will see gradual improvement; in others, we may experience a revolutionary breakthrough. One thing is clear – our life will be more and more influenced by robotic technologies. Microrobotics will play a significant role in this fascinating development.

Further reading: “Automated Nanohandling by Microrobots” (Springer, 2008) and “Mobile Microrobotics” by Metin Sitti (MIT Press, 2017)

Professor in the Department of Computing Science, Head of Division for Microrobotics and Control Engineering at the University of Oldenburg, Germany; Member of the Editorial Board of Jour. of Micro-Bio Robotics.
Did you like it? Share it with your friends!
Published items
0779
To be published soon
+92
New