The original article is titled “The DIY electronics transforming research”
Published on April 5, 2017, in Nature News
Original author: Daniel Cressey
The emergence of cheap, streamlined microcontrollers allows users to meet enormous computing power needs in very small spaces.
In front of a subject, a brush is slowly rubbing a rubber hand on the table; meanwhile, her own hand, out of sight, is receiving the same stimulation. By the end of the experiment, the subject begins to feel that the rubber hand is her own.
This “rubber hand illusion” is just a psychological trick. But it requires careful design: generally, to create such an illusion, the operator needs to practice repeatedly how to simultaneously brush the human hand and the rubber hand. To study how people perceive their bodies, Isa Rao, a PhD student in psychology at the University of Glasgow, followed this design but constructed an automated device to manipulate the brush instead of relying on a trained assistant. The core of this device is an Arduino microcontroller.
by the Project Twins
A large number of cheap, streamlined, and easily modifiable computing devices are changing the DIY electronics landscape, with Arduino being one of them. Moreover, they are increasingly transforming the research community (see the accompanying image “The Rise of Microcomputers”). A basic circuit board can cost as little as 30 RMB, while 400 RMB is enough to purchase a complete set of equipment, including power supply, casing, and wires. These systems often lack fancy features, and their learning curves can sometimes be quite steep, but devices like Arduino and Raspberry Pi have packed considerable computing power into very small circuit boards, providing ample opportunities for automation, networking, and data collection and analysis.
For researchers, these features offer both economic and practical advantages. Users can fit such systems into very small spaces, deploy them without monitors and keyboards, and purchase them in bulk to be installed in automated devices that need to be deployed in remote field survey locations and send back data. Achieving all this requires just a bit of creativity.
Although researchers have been personalizing computers and integrating them into their experiments for decades, the mass popularity of small, cheap “single-board computers” has only emerged in recent years. In 2012, the fully functional Raspberry Pi, capable of running the Linux operating system, was launched; by September 2016, it had sold ten million units.
Since its launch in 2005, Arduino (strictly speaking, it is not a computer but a programmable microcontroller) has also been immensely popular. The ready-made components for these two devices include cameras, motion sensors, thermometers, and Bluetooth adapters. Even the International Space Station has a unit developed for a promotional project; this device is equipped with gyroscopes, accelerometers, magnetometers, and various sensors for measuring temperature, pressure, and humidity.
On Earth, Raspberry Pi, Arduino, and other similar devices are used to build various devices: underwater recorders for marine biology research, robots for assembling gene editing tools, and systems that can quickly identify antibiotic-resistant pathogens from human samples.
Hakho Lee of Harvard University focuses on medical diagnostic instruments; he has developed a system for diagnosing antibiotic-resistant bacterial infections. He said that initially, he tried using a smartphone as the central computer for the instrument, but modifying smartphones is not easy. In contrast, devices like Arduino and Raspberry Pi are designed for modification; integrating them into research sensors is perfectly suitable.
“For this need, Arduino is simply perfect,” he said, “This series has long had various models, and they are all modular. If I want to add Bluetooth functionality, I just need to buy a Bluetooth module and plug it into the motherboard.”
Although Lee himself has extensive experience with electronic devices, such a background is not necessary: the plug-and-play nature of these devices has attracted a large number of researchers without electronic engineering or programming experience. Turning a Raspberry Pi circuit board into a functioning computer is very simple; just insert a pre-installed SD card with the Pi operating system, then connect a monitor, mouse, and power supply. Programming this device requires more work, but for most people, detailed online guides should be more than sufficient.
In fact, the Raspberry Pi was originally intended for teaching programming to children in the UK. Both Raspberry Pi and Arduino use a C++-based language, so having a basic understanding of this programming language helps, but it is not essential for beginners, according to Kenji Takemoto, a researcher at the Kavli Institute for Brain and Mind at the University of California. “Pick a project and start experimenting.” (One example is a variant of the familiar “Hello, world!”; here, the goal is not to display information on the screen but to light up and extinguish the onboard LED.)
Takemoto developed a device called “Flyception,” which is also based on an Arduino; this instrument can wirelessly monitor brain activity in fruit flies in response to stimuli. The Arduino processor controls camera triggers and other functions. “It (Arduino) can synchronize a lot of mechanical and electronic devices with our imaging and behavior recording devices, packaged into a simple and easy-to-use collection, and that is its charm,” he said.
Before researchers start assembling devices, they must first decide what to buy: they are faced with a dizzyingly vast market. Raspberry Pi now has three basic models, along with a $5 “Pi Zero” basic version and a low-power version. Arduino has even more models. In addition, there are Intel’s Galileo and Edison boards, the open-source Banana Pi, and Android-based Odroid, among others.
Paolo De Souza, who conducts research on microsensor technology and devices at the Commonwealth Scientific and Industrial Research Organisation in Australia, says that when choosing devices, you need to carefully consider three attributes: cost, functionality, and size. “Generally speaking, you must give up one of these three; finding a device that is cheap, small, and powerful is not easy.”
For example, De Souza has worked on the development of automated machinery such as submarines and rovers, which require low power consumption and the ability to handle complex sensor data. His team chose the Odroid XU3 (now replaced with the $60 XU4); this device is more expensive than Raspberry Pi but much more powerful. However, in another study on bees, he used Intel’s Edison: it requires less power and can even be powered by solar energy.
Meanwhile, scientists also need to consider how their computers will interact with other elements in the experiment. Raspberry Pi can connect to standard monitors, USB keyboards, and mice. But as a microcontroller, Arduino needs to be connected to a computer via external wires for programming.
Not all microcontrollers have the same capabilities. For example, most devices can read information from a micro USB port, but some devices also have Bluetooth and Wi-Fi capabilities. If you want to build an experimental device that can transmit information to another computer for analysis, or if you need to connect cameras or other external sensors for complex research, these factors become very important. In De Souza’s bee study, the bees were equipped with RFID chips, which triggered corresponding sensors as they flew by, while the Edison computer located inside the hive processed the generated data and uploaded it for analysis.
“Consider your options; you might also want to sort out your needs and see if you can cut a few of them,” De Souza said. As systems are continually updated, you should also be prepared to upgrade your devices: in De Souza’s bee observation project, the single-board computer he initially used was produced by a company called Ledato, then switched to Raspberry Pi, followed by Galileo, and finally migrated to Edison.
For beginners, choosing the most popular devices—namely, Raspberry Pi and Arduino—has its advantages.
The user forums for these two devices are extremely popular, with millions of posts and hundreds of thousands of members, providing very practical resources for newcomers. Researchers often seek assistance on these forums, asking questions such as how to accurately calculate tree circumference with Arduino, or how to set up temperature sensors. The Raspberry Pi development organization even provides a series of videos and common error lists on its website to help newcomers get started.
“What you least want to do is deal with technical problems that others have already solved,” said David White, a computer scientist at University College London, who has extensive experience in Raspberry Pi development. “Using standard solutions can save a lot of time. You can take full advantage of the high popularity of these devices, as there are many successful examples to learn from.”
But obtaining help does not necessarily require relying on the internet. “You can talk to people who have worked with these devices before,” White said. De Souza offered another approach: recruit some graduate students with information technology backgrounds. “IT students are very passionate about their field and always keep up with the latest solutions,” he said.
Some researchers ultimately find that Raspberry Pi or Arduino components are still insufficient to meet their experimental needs. After discovering that Arduino’s system clock was not accurate enough to meet the requirements of the Flyception system, Takemoto switched to the myRIO controller from National Instruments. Lee plans to use NVIDIA’s Jetson to provide higher processing power for future projects.
But for most people, the simple, compact, and inexpensive Raspberry Pi and Arduino are all they need. This is also Rao’s sentiment. “Arduino is truly a gift from heaven,” she said.ⓝ
Nature|doi:10.1038/544125a
Click “Read the original” to read the English original
Related Articles
“Differentiable Neural Computers” can learn like neural networks and process complex data like computers.
Click here to read: What is Google Deepmind’s latest developed differentiable neural computer?
Copyright Notice:
This article has been translated by the Springer Nature Shanghai Office. The Chinese content is for reference only; all content is subject to the English original. Feel free to share it on your social media; for reprints, please email [email protected]. Unauthorized translations are infringing acts, and the copyright holder reserves the right to pursue legal responsibility.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All Rights Reserved
