Category Archives: Newsworthy

Where are AREE Rovers Going?

NASA is planning new types of rover explorers for observing extreme environments, such as the surface of Venus. They plan to build simple yet robust vehicles. AREE is their acronym for Automaton Rovers for Extreme Environments.

NASA’s Curiosity Rover on Mars has been roving and exploring the planet’s surface for the last five years. Among rovers, Curiosity is at the top position. It uses special systems for rejecting heat, X-band receiver and transmitter for communicating directly with Earth, an Electra-Lite radio (UHF) for communicating with the Mars Orbiters, instruments for mineralogy and chemistry, instruments for simple analysis, and much more.

According to Jason Derleth, NASA prefers to do the absolute maximum when sending a rover into space, such as making sure the rover can contribute as much to science as is possible. Jason is the head of NASA’s Innovative Advanced Concepts Program (NIAC).

However, Venus is vastly different from Mars. Although very similar to Earth in its size, mass, and density, Venus has an incredibly thick atmosphere—a mix of carbon dioxide, nitrogen, and sulfur dioxide. This raises the temperature on the surface of Venus to over 450°C, which is hot enough to melt lead or high enough for paper to spontaneously combust. The atmospheric pressure at the surface is 92 bar or 1,334 psi, with a density enough to crush a submarine.

In the past, some robots have succeeded in reaching Venus. These were the Soviet Union’s Venera and Vega landers, and the Pioneer probe from NASA. Although they were successful in reaching the planet’s surface, they could function only between 23 and 127 minutes before the oppressive environment snuffed out their electronics.

With the AREE rovers, NASA is trying a new concept, inspired by mechanical clockwork computers and tanks used in World War I. A NASA program, NIAC, is funding the AREE rovers. It is offering small grants for developing early stage technology, which allows engineers to work on long-term ideas for properly developing the technology.

For instance, the most recent funding from NASA related to the development of a rough prototype of the rover concept, which will take about three years. Jonathan Sauder was the first to propose the concept. In 2015, Sauder had observed mechanical computers using levers and gears for performing calculations rather than rely on electronics.

The AREE rovers would be using these analog techniques mainly to survive the harsh environments on Venus. They would traverse the planet’s surface moving on tank treads that overcame the rough terrain. As wind gusts on Venus are high, they would turn wind turbines located at the center of the rover to supply it with the necessary power. The robot would capture the power from the turbines inside springs before distributing it to the other subsystems of the robot. Think of a windup watch, the idea is very similar.

Curiosity has several cameras to measure, map, and guide it over the Marian terrain. However, the electronic functionality of the AREE rovers will be purposely kept simple. Although AREE’s design will make it robust enough to withstand unexpected bumps and drops, it will integrate a simple optical reflector to transmit data to the orbiting satellite.

Butterfly IQ – Smartphone Connected Ultrasound Scanner

Traditional ultrasound scanners are rather expensive, and rarely do people own one to use at home. However, that may be about to change, as Butterfly IQ has now obtained FDA clearance for a portable ultrasound scanner that anyone can use by connecting to their smartphones.

Connecticut-based Butterfly IQ has made an innovative ultrasound scanner that uses a semiconductor chip for generating the ultrasonic signals, rather than the piezoelectric crystal transducers that traditional ultrasound machines use. The semiconductor chip based transducer is much easier to manufacture than the piezoelectric ones are.

Using the semiconductor chip makes the device much less expensive as compared to existing ultrasonic scanners. The cost of ownership comes down further as the device can operate with a smartphone and other smartphone connected devices such as the Philips Lumify device.

According to Dr. Jonathan Rothberg, founder and chairperson of Butterfly Network, this ultrasound-on-a-chip technology opens up a low-cost window for peering into the human body, allowing anyone to access high quality diagnostic imaging. With more than two-third of the population of the world without access to proper medical imaging, this effort by Butterfly is a great beginning.

FDA has cleared the device for 13 different clinical use cases. These include pediatric, urological, gynecological, cardiac, abdominal, and fetal use cases. The scanner transfers the captured imagery directly to the user’s smartphone via a chord, and the smartphone stores the images into a HIPAA-compliant cloud.

As reported by the MIT Tech Review, the chief medical officer of Butterfly Network, Dr. John Martin, was able to detect cancerous growth in his body while testing the scanner. This is an example of the potential of the low-cost ultrasound scanner.

According to Martin, the easy-to-use, powerful, healthcare providers will be able to afford the whole-body medical imaging system for less than $2,000, and it will fit in their pockets. As the price barrier comes down, Martin expects the Butterfly device to replace the stethoscope ultimately in the daily practice of medicine. The impact this technology will provide as a low-cost diagnostic system, can be gaged from the help it will offer to hundreds of thousands of women who die in childbirth, and the millions of children who die of pneumonia each year.

After perfecting the scanner, Butterfly has plans to augment its hardware capabilities with software for artificial intelligence. This will help clinicians interpret the images that the device picks up. The company expects the products with many new features to be ready for the market by 2018. At present, the device works only with iPhones.

According to the President of Butterfly IQ, Gioel Molinari, ultrasound imaging makes a perfect combination with deep learning. With more physicians using the devices in the field, the neural network models keep improving. As physicians use the Butterfly scanner regularly, they will be able to interpret the results better. This will help improve the acquiring and interpretation of the image by the artificial intelligence, which in turn, will help less skilled users to extract life-saving insight from the images captured by the Butterfly IQ ultrasound scanner on the field.

Progress in the World of Internet of Things

Although many in the electronics field lambast the Internet of Things (IoT) as an inappropriate or inadequate acronym, IoT is a space to huge to be confined to these narrow adjectives. In reality, IoT requires a blanket description, as it covers a vast arena. Problems arise from compartmentalization and although various spaces such as industrial and medical have established a big head start, others have yet to launch their true separate identities.

For the electronics designer this means taking the general palette of IoT features and functionalities and tailoring them specifically to the application at hand. The designer must be knowledgeable about state-of-the-art technologies such as those required for cloud connectivity, wireless design infrastructure, interface ergonomics, and internal power management. The designer must be familiar with the methods of manifesting them in their design, as these may be critical aspects.

For instance, there are several suggestions for scaling the IoT from smart factories to smart homes. Although there are blueprints for pollution reduction, city traffic management, and electrical energy distribution, the purveyors of industrial-grade operating systems do not yet have a detailed plan for the smart home.

According to Wei Tong, Product Marketing Manager of Dialog Semiconductors, wearable technologies can do far more than simply functioning as personal items. Using Bluetooth, a communications standard protocol, wearable devices can connect to a larger network, allowing them to communicate with other devices via beacons and sensors, thereby manifesting the larger Internet of Things.

However, despite the birth of the phrase “the Internet of Things” 18 years ago, and the first connected IoT device 35 years ago, consumers are yet to adopt wearable IoT in mass quantities. According to Nick Davis, this is due to two factors—first, ease of use, or lack thereof, and second, lacking the purpose or serving the wrong purpose.

For instance, take the case of “smart” light bulbs. Some are easy to connect to and control with smartphones, while others give users a hard time. According to Nick Davis, once people face such difficulties, they tend to give up on the entire IoT and smart device concept.

Another example Nick Davis gives is that of a smart toaster or smart refrigerator and the purpose they serve. According to Nick, most companies have not done proper market research into the actual requirement of people who use toasters and refrigerators, and what the consumers expect in such smart devices. However, several new products on the market are potentially useful to designers.

Another example of wrong purpose is the video sunglasses from Snap, the parent company of Snapchat. These are basic sunglasses with a video camera attached. They allow users to capture and post videos more easily to Snapchat. According to Nick, Snap is stuck with hundreds of thousands of their unsold spectacles. Apparently, Snap did not realize that people are not very keen on walking around taking videos with their eyewear.

Despite such debacles above, newer products are appearing on the market that help designers achieve better energy-efficient IoT products, voice recognition engines, and flexible and smart motor-control options that are also lightweight and compact.

Why Panic Buttons are Going Wireless

Panic buttons or emergency stop switches are extremely important for protecting workers, machinery, and products from catastrophic failures. Traditionally, manufacturers include them with their machinery, and most are hard-wired. However, things are changing and now, these red emergency switches are finally going wireless.

When a machine malfunctions, or a critical incident occurs, operators often have to press these last resort switches to bring the system or the entire machine to a halt quickly and safely. Hence, these switches are aptly called E-stop, emergency, or panic switches. Operating these switches brings the machine or the system to a halt and prevents serious damage to products or the machine itself, as well as preventing injury to workers.

The importance of the emergency stop button is evident from a report from OSHA or Occupational Safety and Health Organization. According to this report, more than 5000 US workers were injured fatally in 2015 in industrial accidents.

Ever since the second industrial revolution, manufacturers have hard-wired E-stops in their machines as a standard solution to shut them down in case of emergency. Usually, manufacturers placed these emergency switches well apart from the usual on/off and other switches the machines normally carry on their control panels, making it easier for the operator to identify and hit them to stop the machine. With the E-switches being functionally so important, it is understandable manufacturers were reluctant to make them wireless. However, a wireless E-stop device would allow the worker to shut the machine down without even having to go near it, improving the safety factor.

The tech company, Laird PLC, of London, has seemingly realized the additional benefit of a wireless E-stop button, and has evolved the Safe-E-Stop. It is possible to incorporate the Safe-E-Stop with the existing hard-wired emergency stop system already involved with production systems such as assembly lines. This improves the on-the-job safety, as an individual operator or a group can immediately shut down a machine in the production line, without having to hit the hard-wired E-stop button physically.

The emergency might involve the closest machine-mounted E-stop button in the same danger zone. Therefore, the operator rushing in to operate such an emergency button could face a hazard and increase the response time for arresting the emergency.

Laird PLC has developed a wireless personal safety solution rated at SIL 3 as an answer to the above problem. The Rockwell Automation distributors market the Safe-E-Stop from Laird making it available through the Encompass partner program of Rockwell Automation.

Users can have continuous status indication on LED and LED readouts on the Safe-E-Stop system. They can use the IP/Ethernet port on the MSD or Machine Safety Device for reporting the status of the wireless E-stops actuated to personnel in charge of operations. It is possible to link as many as five PSDs or Personal Safety Devices to the MSD simultaneously. This allows multiple operators to collaborate or work independently to supervise the operation. Activation of an E-stop on any linked PSD causes the MSD to issue a stop command and notify all other PSDs immediately of the stop condition.

Beemo Works With Raspberry Pi

Even adults watching Adventure Time wish to own a personal BMO, the quirky living game system from the Be More episode of the show. Although based on the GameBoy, BMO is a digital friend calling out through the nostalgia lens of our childhood times. Now Bob Herzberg has created Beemo, a BMO for his daughter and her friends.

In building the living little boy, Beemo, Herzberg used the popular single board computer Raspberry Pi (RBPi), which be runs on battery power, a USB battery pack. Although his body is made from laser-cut MDF wood, Beemo uses am 8-inch HDMI monitor. Herzberg had to 3-D print the arms and legs, attached them to the body, which he sanded, sealed, and painted. Adding some vinyl lettering completed the look. Adding a small wireless keyboard meant Beemo could be remotely controlled.

To interface the gaming button on the panel, Herzberg had to create a custom PCB, and he laser-cut the special acrylic buttons to mount them. These he connected to the IO header on the RBPi to make them work. Another PCB functions as a holder for the USB sockets. This allows Beemo to have USB ports on the front panel. Beemo works comfortably for a continuous 8-hour period on his battery.

Herzberg’s daughter created the custom animations that he then transformed into MP4 video files—giving Beemo most of his personality. The remote keyboard operations turn the animations on. Some BMOs are given an internal microphone and a speaker. The BMO translates the user’s voice using Google Voice API, and maps it to an appropriate response, allowing the user to have a conversion with BMO.

Herzberg also used the RBPi camera module. Some BMO makers use servos to make the camera pop out for taking a snap. This type is called the GoMO and it can take pictures. Actually, there is a whole family of MOs—GoMO, CMO, XMO, UMO, and a few others. Although people like to think of the retractable camera as a ghost detecting equipment, Beemo simply likes to take nice photos.

Playing games with Beemo is very simple. You only have to load one of the emulators Raspbian supports. Raspbian is the operating system that makes RBPi run. Herzberg faced some real challenges when creating Beemo. He had to use different materials and techniques to fabricate the enclosure. However, the presence of the RBPi inside meant bringing Beemo to life was much simpler.

While Beemo may not be able to hop around and sing as the BMO in Adventure Time did, he can certainly play a huge number of retro games, because of the RBPi within him. As Herzberg was familiar with the Atari 800 emulator, having written games for that platform earlier, he used the front panel USB ports for connecting gamepads. Of course, the D-pad and front panel buttons are also equally useable.

Herzberg uses the RBPi A+ as the heart of his project. He has split 256 MB of the RAM between the CPU and the GPU. He also uses the composite video and stereo outputs on the 4-pole jack internally. By modifying the config.txt file, he was able to shut off HDMI output completely.

What is E-Smog and How to Detect it?

Many people claim advancement in technology and the proliferation of electronic devices is creating a sea of electromagnetic waves around us, and this eSmog is actually a cause for many of the illnesses we are afflicted with nowadays. While eSmog causing bad health is up for debate, some people seem to be more sensitive to it than others are. However, the presence of electromagnetic waves around us cannot be ruled out, with greater concentrations around devices such as computers, mobile phones, Wi-Fi routers, cordless phone bases, and in fact, anything electronic and powered up. Therefore, an instrument that measures the level of electromagnetic fields around it is in order.

Today, it is practically impossible for us to live life without our electronic devices and everyday technology that produce electromagnetic fields. Although we cannot see the electromagnetic fields that surround us, an instrument that can measure its presence is useful for us to know, say, whether a brick wall has reduced the level, and to what extent.

We all need our Wi-Fi, Zigbee, Bluetooth, television, radio, mobile phones, and other gadgets. To know the level of eSmog each of them is producing, you can use the kit TAPIR—an eSmog detector. You can assemble this tiny instrument from the seven small PCBs in the kit. TAPIR comes with an antenna and two types of electromagnetic detectors.

The kit has a PCB panel, actually made of seven parts. You can assemble the PCBs and make them form the enclosure for TAPIR. The PCBs are numbered—starting with the top piece, the left sidepiece with a switch, the bottom piece with the components, the right side piece with the headset connector, the negative battery connection piece, the positive battery connection piece, and the end piece. A headset plugged into the connector allows the user to hear the device detecting eSmog. The intensity of sound increases with the level of eSmog TAPIR detects. You need a single AAA battery to power the kit.

TAPIR—acronym for Totally Archaic but Practical Interceptor of Radiation—is a wideband ultrasensitive eSmog detector. Once you have connected it to the antenna and the headphones, and switched it on, you can move it around an electronic device. This allows you to hear different noises depending on the type and frequency of the field the device is emitting.

Making the two antennae for the TAPIR is important for it to function properly. All around us, there are two types of electromagnetic fields—the E-field or electrical field, and the H-field or the magnetic field—and two separate antennae are necessary to allow TAPIR to detect the two fields.

The E-field antenna consists of a length of solid insulated wire. The kit includes the wire, and you will need only half of it to form the antenna. Insulate one end of the wire with heat-shrink tubing and bend it to form a loop. At the other end of the wire, solder the cinch connector shell to complete the antenna.

A coil is enclosed with the kit, and you can solder this coil to two pieces of insulated wires. Solder the free ends of the wires to the second cinch connector, and your H-field antenna is ready.

Accurate Power Monitoring with LTC2992

Linear Technology Corporation, now a part of Analog Devices, Inc., has recently placed on the market a power monitoring IC, LTC2992, which offers a wide-range, dual monitoring system for current, voltage, and power for 0-100 VDC rails. The IC is self-contained and does not need additional circuitry for functioning.

Users get a variety of options for operating the LTC2992. For instance, they can derive power from a 3-100 VDC monitored supply, or from a 2.7-100 VDC secondary supply, or from the shunt regulator on-board. Therefore, when monitoring the 0-100 VDC rail, the designer does not have to provide a separate buck regulator, a shunt regulator, or an inefficient resistive divider.

Within the LTC2992 are a multiplier and three Analog to Digital Converters (ADCs) of the delta-sigma type. Two of the ADCs provide measurements for current in each supply, while the third ADC measures voltage in 8- or 12-bit resolution and power in 24-bit resolution. The wide operating range of the LTC2992 makes it an ideal IC for several applications such as blade servers, advanced mezzanine cards, and 48 V telecom equipment.

Users with equipment using negative supply or supply greater than 100 VDC can make use of the onboard shunt regulator. The LTC2992 has registers that one can access with the I2C bus, and it uses these registers to store the measured values. It can measure current and voltage on-demand or continuously, using these to calculate the power, and stores this information along with maximum and minimum values in the registers.

The LTC2992 has four GPIO pins, which the user can configure as ADC inputs for measuring neighboring auxiliary voltages. Over its entire temperature range, the LTC2992 takes measurements with only ±0.3% of the Total Unadjusted Error (TUE). For any parameter going beyond the thresholds programmed by the user, the LTC2992 raises an alert flag in the specified register and on the specified pin. This is according to the alert response protocol of the SMBus.

The I2C bus on the LTC2992 operates at 400 kHz and features nine device addresses, a reset timer for a stuck bus, and a split SDA pin for simplifying the opto-isolation for the I2C. Another version of the IC, the LTC2992-1 offers users an inverted data output pin for the I2C. This makes it easy for the users to interface the IC where the opto-isolator has an inverting configuration.

The ICs, LTC2992 and LTC2992-1, are both available in automotive, industrial, and commercial versions. Their operating temperature ranges are -40°C to 125°C for automotive, -40°C to 85°C for industrial, and 0°C to 70°C for commercial applications. Linear Technology Corporation makes both versions of the IC in packages of 16-lead MSOP and 16-lead 4 x 3 mm DFN, and both versions are RoHS-compliant.

Most electronic applications require monitoring of current, voltage, and power at board level. Knowing the key system parameters provides valuable feedback, allowing users to monitor the health of their systems and make intelligent decisions. They help in determining whether a system is operating properly, efficiently, or even dangerously. Users can choose for various types of monitoring ICs, ranging from hot-swap dedicated power ICs to temperature monitors.

It is Time for Chip Speakers

So far, speakers have been electromechanical devices, with a coil moving within a magnetic core, attached to a baffle or driver to move the air for producing the sound. With devices going down in size, manufacturers have been facing difficulties in producing electromechanical speakers in smaller sizes. Piezoelectric speakers are available, but they operate on a very narrow bandwidth.

Now USound GmBH, from Graz, Austria, has presented an audio speaker based on micro-electro-mechanical-system (MEMS) technology. This chip-sized speaker is suitable for small equipment such as Internet of Things (IoT) devices, wearables, smartphones, and earbuds.

By the end of the current year, USound expect to reveal Megaclite, a reference design using its MEMS speaker, Ganymede. So far, USound has fitted Ganymede to sunglasses at the high end. According to USound, Ganymede is suitable for mobiles, earbuds, and high fidelity, multidriver speakers playing above ear levels.

According to Mark Laich, senior adviser for business development at USound, making the diminutive MEMS drivers sound good across the audible spectrum was a huge challenge for the engineers. The major difficulty they faced was from the sound related physics, as it dictates the diaphragm size to push the air to be proportional to the wavelength of the sound emitted. That is why high-fidelity speaker systems use 12- to 15-inch drivers for producing low frequency bass sounds, 3- to 6-inch midrange drivers for the mid-frequency sounds, and 1 or less than 1-inch tweeter speakers for producing high-frequency sounds.

For the tiny speakers used in wearables, the size of the driver has to be some small portion of the wavelength of the sound it emits. Usually, some electronic or mechanical frequency equalization is necessary to make them sound high fidelity. Highest fidelity, as some headphones at the high end provide, is achievable only with multiple drivers. Typically, most of the reasonably priced earbuds have to sacrifice fidelity as they use a single driver, while adding electronic equalization to sound better.

As it is not possible to circumvent the sound related physics, MEMS speakers from USound are similar. Their low-end model has a single driver along with electronic equalization within a chip-scale package, and this bonds directly to the MEMs die. The MEMS frame is actually a longish actuator that moves a diaphragm using suspension beams made of piezoelectric material. The surrounding diaphragm also seals the entire chamber.

According to Laich, this arrangement achieves high-speed actuation, with a response time in microseconds. The company says this will help in noise cancellation in models to come, when they build them with a MEMS codec partner. At present, the air-pushing cone or diaphragm lies at the bottom side of a cavity, with thin piezoelectric drivers suspending it by the corners. The drivers supply the necessary energy to move the diaphragm in synchronization with the audio signal.

Listeners describe the sound from the MEMS speakers as digital, similar to the sound from a CD in comparison to that from a vinyl record. Of course, even when fortified with electronic equalization boosting the low frequencies, the sound from a single driver design does not match the high fidelity demonstrated by multidriver design.

Oscilloscopes Lose their Faces

The word oscilloscope usually conjures up images of a box with a display. Earlier, oscilloscopes were bulky devices with a display made of a cathode ray tube, but later the models became sleeker, and came with a liquid crystal display. Another difference was in their method of measurement. Whereas there were analog units earlier, later models sported an analog to digital converter inside, which converted all analog signals to digital data. Nevertheless, the display continued to be a part of the oscilloscope.

However, Tektronix now has unveiled a low-profile oscilloscope that has lost its face—the display. The MSO5 series of oscilloscopes from Tektronix has a faceless version aimed at Automated Test Equipment (ATE) applications. It is a low-profile version competing with modular oscilloscopes and digitizers.

Suitable for automated tests or for monitoring machines, the low-profile MSO58 of the MSO5 series, is a rack mountable unit. All its specifications match those of its regular benchtop cousins. It has eight analog inputs with FlexChannel features, which allow eight digital channels to substitute the analog channel with a 1-GHz bandwidth on all of them. The real-time scan rate for all the channels is 6.25 Gsamples/sec with 12-bit ADCs on each channel, but a high-resolution mode allows the resolution to increase to 16 bits and 125 Msamples/sec. That makes the effective number of bits as 7.6 at 1 GHz, or 8.9 at 20 MHz, with a record length of 125 Msamples/channel.

Software within the faceless oscilloscope can help with jitter and serial bus analysis, channel math and Fast Fourier Transformations (FFT). For bench and debugging applications, the software also provides cursors. There are six USB host inputs, on USB input for a device, a LAN port, a Display Port, DVI-D port, SVGA output port. However, the device lacks GPIB connectivity.

As the internal hardware and functionality is identical in both the benchtop and the faceless versions of the MSO5 series oscilloscopes, any automation code for production for device validation and characterization works interchangeably. The six USB host ports may lead one to believe the low-profile oscilloscope could be useful as a system controller. However, the operating system of the unit is a closed Linux version, and a separate PC is necessary for automated use.

The six USB device ports can help in creating a network, to which, one can add more accessories such as an external storage or other instruments. If you have additional MSO5 low-profile units to work together, you can also add a USB hub or an Ethernet switch. Unfortunately, for those using GPIB primarily, these units do not come with a GPIB port.

It is very easy to configure any input of the low-profile faceless oscilloscope as one analog or 16 logic channels. Therefore, one can mix and match the configuration to change it as necessary. For instance, channels 1 and 2 can be analog, while the channel 3 caters to 16 logic inputs.

Bandwidths for the MSO5 series oscilloscopes are 350 MHz, 500 MHz, 1GHz, and 2 GHz. However, one can upgrade any model at any time to operate at any bandwidth.

Facial and Object Recognition with A Raspberry Pi

f you are using the single board computer Raspberry Pi (RBPi) for vision-related tasks such as facial and object recognition, the NCS or Movidius Neural Compute Stick from Intel could help to boost the rate at which the RBPi carries out its tasks—you actually do not need to employ a server farm for the job.

The RBPi is fully capable of running software for facial image recognition, and hobbyists have long being using the SBC for recognizing faces in videos to identifying obstacles in the path of a robot. However, the rate at which the RBPi carries out such tasks leaves much to be desired, and the NCS helps to improve this rate.

The Movidius NCS from Intel plugs into the RBPi via the USB port. Inside the stick is a Myriad 2 Vision Processing Unit (VDU) with 12 specialized cores that accelerate the vision recognition tasks for the RBPi. Although it consumes only a single watt of power, the low-power VDU processor works at 100 gigaflops. Sometimes, the stick may need higher processing power and it could consume 2.5 W.

Users can watch the video Movidius has released for guidance on how to use the NCS. There is also a text guide to help users figure out the nuances of object recognition using the RBPi and the NCS. The video demonstrates the system recognizing a pair of sunglasses and a computer mouse on the table.

To get the demo running, the user needs to download and install a few software libraries. On the hardware side, apart from the RBPI, you also need a Pi camera.

Movidius initially announced the early version of the NCS in April last. They then released a prototype device, which they named Fathom, before Intel purchased Movidius. According to Dr. Yann LeCun, founding father of Convolutional Neural Networks, and director of AI research at Facebook, Fathom was a significant step forward.

Intel then released NCS, which has broadly the same specifications as the Fathom did, with the exception that the former has a 4 GB memory. This is an improvement of four times over that of the latter, and it helps the NCS to support denser neural networks. With NCS, any robot, big or small, can possess vision capabilities that are state-of-the-art.

According to Intel, the NCS can lower the barriers for those starting with deep learning application development. It actually offers a simple way for users to add a visual recognition system to their prototype devices such as robots, surveillance cameras, and drones.

As the NCS already has 4 GB of internal memory, and handles all the data in a neural network that is locally stored, the NCS does not have to rely on an Internet connection to connect to a server. In actual practice, transferring data to and from a remote server would introduce a huge latency and any high-performing processor to overcome the latency would consume a huge amount of power. The NCS overcomes both the above shortcomings.

The processor on the NCS is more powerful than the RBPi, although it does not actually accelerate the training process of a neural network, which is a computationally intensive process when carrying out vision recognition.