Reliable systems for micro aerial vehicles — MAV challenges

Editor's Note: Embedded designers must contend with a host of challenges in creating systems for harsh environments. Harsh environments present unique characteristics not only in terms of temperature extremes but also in areas including availability, security, very limited power budget, and more. In Rugged Embedded Systems, the authors present a series of papers by experts in each of the areas that can present unusually demanding requirements. A separate excerpt of the book addresses fundamental concerns in reliability and system resiliency.

This excerpt describes specific requirements and approaches for addressing reliability in unmanned micro aerial vehicles (MAVs) presented in the following installments:

Elsevier is offering this and other engineering books at a 30% discount. To use this discount, click here and use code ENGIN317 during checkout.

Adapted from Rugged Embedded Systems, Computing in Harsh Environments, by Augusto Vega. Pradip Bose, Alper Buyuktosunoglu.

CHAPTER 7. Reliable electrical systems for micro aerial vehicles and insect-scale robots: Challenges and progress
By X. Zhang, Washington University, St. Louis, MO, United States

The dream to fly has sent men on an ardent pursuit to build ever more sophisticated aerial vehicles. Over the years, steady progress has been made to miniaturize such systems and has led to prolific development and maturation of drones. Recent research advancement endeavors to drive the miniaturization further to the micro scale, and promising prototypes are being developed to explore this exciting yet untrodden frontier. In this chapter, we present the progress our research team has made on building an insect-scale aerial vehicle and zoom into the critical reliability issues associated with this system. Since many such systems are envisioned to operate in harsh environment hostile or harmful to human beings and have stringent design requirements and operation constraints, reliability becomes a first-class design consideration for these microrobotic systems.

We will first give a brief background on micro aerial vehicles (MAV), the opportunities they offer, and the challenges they present, followed by an introductory overview on the insect-scale MAV prototype called RobeBee, which is currently under our development to demonstrate autonomous flight. Next, we will dive into the detailed implementation of its electronic control system and elaborate the reliability concerns and considerations in our design. Improved performance and reliability have been confirmed in both simulation and experiment after applying codesign strategy between supply regulation and clock generation. Finally, we will conclude with our vision on the future of MAV from a reliability perspective.



Unmanned aerial vehicle technology has indeed taken off and its tremendous commercial success and wide adoption in many fields has also fueled increasing recent interest in MAV, which loosely refer to air craft with size less than 15 cm in length, width, or height and weigh less than 100 g. These systems are envisioned for applications including reconnaissance, hazardous environment exploration, and search-and-rescue, and therefore may require various morphologies that can be broken into a number of classes such as fixed wing, flapping wing, or rotary wing.

The history of MAV dates back to 1997, when the United States Defense Advanced Research Projects Agency (DARPA) announced its “micro air vehicle” program [1]. The technology advances that propelled this bold marching step are the maturation of microsensors, the rapid evolution of microelectromechanical system, also known as MEMS, as well as the continued exponential improvement of computing technology. In 2005, DARPA again pushed the limits of aerial robotics by announcing its “nano air vehicle” program with tighter requirements of 10 g or less and within 7.5 cm dimension. These programs have led to successful MAV prototypes including Microbat [2], Nano Hummingbird [3], and inspired a number of recent commercially available flapping-wing toy ornithopters and RC helicopters on the scale of MAVs.

Taking the pursuit of miniaturization to the next level, researchers are now working on “pico” air vehicles that have a maximum takeoff mass of 500 mg or less and maximum dimension of 5 cm or less [4]. As this size and weight range falls into the scale of most flying insects, it is no surprise that many successful MAV systems developed at this scale are modeled after insects. The Harvard RoboBee project, which is the focus of this chapter, is one example prototype of a “pico” air vehicle.

2.2 WHY MAV?

Why are we interested in the extreme art of building insect-scale MAVs? Rodney Brooks, a renowned roboticist, has put it quite eloquently in his forward-looking paper in 1989, titled “Fast, Cheap, and Out of Control: A Robot Invasion of the Solar System” [5], where he laid out the benefits of employing small robots for space exploration. Fast forward 25 years, rapid development in information technology and advanced manufacturing has pushed the applications of robots well beyond the space exploration missions, yet many of the same benefits Brooks argued in his seminal work remain:

Fast development and deployment time: it used to be that “big” complex systems take years of planning to take shape and the turnaround time to discover any critical problems in the system design can be prohibitive. Similarly, even after the system has been designed, developed, and debugged, its deployment can be equally time-consuming because of its complexity and the intricacies involved to interface it with other complex systems and humans. Therefore, only huge organizations, such as the military and government agencies, could afford the resources and man power required for such missions over an extended period of time, which severely limited the adoption of the traditional large-scale robotic technology. As we miniaturize robots to the microscale by leveraging established design methodology from the IC industry and development practices from software engineering, the time it takes from conception to implementation of the robotic system can be drastically reduced, resulting in faster design iterations and ultimately superior system performance and reliability.

Cheap prototyping and manufacturing cost: if you watched the latest DARPA Robotics Challenge like I did, you are probably blown away with awe too by the bipedal humanoid robot called Atlas that stands 6 foot tall and weighs 330 pounds. You might be asking the same question—“Why can’t I get that?!” The answer probably has most to do with its whopping price tag of $500,000. It is obvious that while robotic technology has been making steady progress in the military and industrial setting, the cost of a large-scale system makes it hard for the same technology to spill over to our everyday life. However, things are rapidly changing in the MAV arena. Moore’s law and its counterpart in sensors have driven high-performance embedded processors and high-resolution cameras as inexpensive commodities; 3D printers, laser cutters, and computer numerical control (CNC) machines have enabled desktop fabrication; and the open source and open standard movements have built a vibrant and resourceful community that is ripe for innovating. Once again, miniaturization has torn down the cost barrier of traditional robots and is poised to release the technology’s full potential.

Collective power of the swarm: one tiny aerial robot that can perform simple tasks may not seem much at first glance, but if you put large number of these autonomous agents together in a coordinated way, they may be more effective and capable than a few large complex individual robots. For example, consider a search-and-rescue scenario. A rescue worker could release a box of 1000 micro aerial robots, each weighing less than 1 g, at the site of a natural disaster to search for heat, sound, or exhaled carbon dioxide signature of survivors. Many of them may fail, but if only a few of these robots succeed, they would have accomplished their mission with significant cost and deployment advantages over current generation of $100,000 rescue robots. This indeed highlights another interesting aspect of MAV reliability that Brooks emphasized—“…that large numbers of robots can change the trade-off between reliability of individual components and overall mission success.”

In addition to the three key advantages offered by miniature robots in general as listed above, an insect-scale MAV can serve as a scientific platform for studying flight mechanics and flight control. Shaped and tuned by nature over millions of years of evolution, insects exhibit unmatched agility and prowess in their flying ability. There are many mysteries yet to be revealed regarding the aerodynamics, the sensorimotor coordination, and the control strategy involved in flight at such small scale. The kind of MAVs we are interested in provides the much-needed alternative methods and tools to investigate these open scientific questions in a controlled, repeatable, and efficient manner.


Miniaturization of autonomous aerial vehicles presents unique technological challenges, especially as we shrink down to the “pico” aerial vehicle scale [4].

First, fluid mechanics changes as a function of characteristic length and velocity. Bird-sized MAVs with Reynolds number (Re) larger than 10,000 that is envisioned in the original DARPA program exist in a regime of turbulent flow and steady lift to drag ratios greater than 10, while “nano air vehicle” (1000 < Re < 10,000) may start to be affected by the impact of boundary layer separation. For insect-scale MAVs (Re<3000), unsteady mechanism beyond constant velocity such as wing flapping can be employed to generate lift, because the flow is almost entirely laminar. However, the energetic cost increases with smaller characteristic length, which penalize pico air vehicles significantly on its flight time. Whereas a larger-scale aircraft may stay aloft based on passive stability for hours or even days, flight time for MAV is expected to be on the order of minutes.

Second, the scaling trend for device manufacturing can be another hurdle. Feature size and characteristic size scale in tandem in aerial vehicle fabrication and assembly. The former refers to the smallest dimension of the mechanical components of the system, such as gear teeth pitch, constituent material thickness, and flexure length, and the latter describes the overall size of the vehicle and thus often relates to wingspan and chord length. As we approach manufacturing of MAVs at the insect scale, the standard machining and assembly tools and “off-the-shelf” components no longer apply, and novel methods have to be developed from scratch. For example, high-resolution CNC mills with positioning accuracy down to 1 μm require rare end mills below 100 μm that are hard to come by. Furthermore, area-dependent forces, such as friction, electrostatic, and van der Waals force become dominant as feature size shrinks, degrading the loss in more traditional bearing joints with respect to power transmission. And the same trend is true with transducers. Friction losses and current density limits worsen the effectiveness of electromagnetic motors with diminishing feature size. Although at first glance, it may seem that MEMS surface micromachining techniques could be the solution to achieve micron-order feature sizes, such methods are often hindered by the time-consuming serial process steps, limited three dimensional capabilities, and high prototyping cost using specialized MEMS foundries. Therefore insect-scale MAVs require innovative solutions to device fabrication and assembly.

Finally, flight control experiences distinctive transitions from larger-scale aerial vehicles to microscale ones. The former can often take advantages of passive stability mechanisms such as positive wing dihedral and enjoy larger mass and power capacity to accommodate various sensors and processors, whereas the latter must work under significantly reduced payload capacity and wrestle for the stringent share of size, weight, and power budget. Therefore, the control challenge in pico aerial vehicles is centered on flight stabilization using constrained sensing and computation capabilities, instead of higher-level control problems such as autonomous navigation and multiple vehicle coordination.

The design and development of MAVs present exciting application prospects, as well as challenging engineering and scientific questions that pique our research interest. In the reminder of this chapter, we will dive into the RoboBee system as a case study of an insect-scale MAV to discuss various design strategies and considerations in building a miniature autonomous flying robot.

The next installment from this chapter describes motivations and approaches taken with the RoboBee system.

Reprinted with permission from Elsevier/Morgan Kaufmann, Copyright © 2016

Professor Zhang joined the faculty at Washington University in St. Louis in 2015. Previously, she was a postdoctoral fellow in computer science at Harvard University, where she worked on the RoboBee BrainSoC and energy-efficient computing projects. She has worked as a graduate research assistant at Cornell University studying variability-tolerant circuits. Zhang earned a doctorate in electrical and computer engineering at Cornell University in 2012. She earned a bachelor’s degree in electrical engineering at Tsinghua University in Beijing in 2006.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.