Aaeon’s “Atlas” is a smart streetlight computer based on its Apollo Lake based NanoCOM-APL module with an optional UP Core Plus SBC. Camera, wireless, and sensors are mated with analytics supplied by Intel’s Movidius Myriad X VPU and OpenVINO toolkit.

Aaeon and Intel announced a fanless, AI-enhanced “Atlas” embedded computer designed to be integrated with a variety of different streetlamp designs. No OS was listed for the “edge node,” which is based on an Intel Apollo Lake SoC and a Myriad X VPU, but we would be surprised if Linux was not supported.

Atlas with streetlamp
(click image to enlarge)

Applications include smart lighting, smart street parking assistance, traffic enforcement, accident response, traffic and congestion monitoring, environmental and seismic monitoring, signage control, real-time traffic signal optimization, license plate recognition, pedestrian crowd control, and flow monitoring applications.

Some of that sounds pretty creepy. Yet, despite the potential for invasion of privacy, there are also numerous benefits ranging from safety to convenience to the big one: improved energy efficiency. Can we build smart cities that work more fluidly and efficiently without at the same time infringing on our privacy and personal freedom? We would certainly hope so.

Intel is handling most of the Atlas marketing as a showcase for its Intel Vision Products line. This includes its dual Movidius Myriad X VPUs, deployed via an Aaeon AI Core XM 2280 M.2 module. The other key element is Intel’s OpenVINO AI toolkit for running AI inferences on frameworks such as TensorFlow or Caffe.

A pair of Intel product briefs outline multiple smart city scenarios such as a smart street parking application. Here, an OpenVINO-based NVR management application streams the Atlas streetlight camera past the Myriad X for real-time analysis before saving it on an SSD or streaming to the cloud via GbE or LTE for in-depth analysis. A license plate recognition application extension automatically charges the driver based on parking duration as determined by the visual analytics. The application would likely be able to process congestion-based parking rates based on time of day and location.

There’s a map visualization integration function, so we can also envision the automatic generation of a real-time online map to enable drivers to find parking spaces. This could reduce congestion, automotive power use, and frustration.

Two views of Edge-X connected streetlight application
(click images to enlarge)

The platform features a cloud-connected smart street lighting application called Edge-X that controls lighting intensity based on ambient light to save on energy costs. Light intensity can be remotely controlled for increasing light in high crime areas — that is if you don’t already have one of those “artificial moon” satellites in geostationary orbit to turn night into day. The application can also identify disabled streetlights and either turn off power to a faulty unit while waiting for a repair team — or in some cases repairing the device remotely.

Aaeon and Intel envision multiple applications running on the same edge node to ease the cost burden of putting a computer and camera on every street corner. In a promotional video, Intel suggests Atlas could more than pay for itself in lower energy costs, but that would likely presume more expensive electricity. The partners are pitching Atlas primarily at governments and companies that are already planning to replace their incandescent streetlamps with more power-efficient LED models.

Value-added functions that might piggyback on the lighting and traffic control system might include real-time air quality reports, alerts for overflowing trash receptacles and water-main breaks, and revenue generation from signage. There’s even something called an “advertising camera” for collecting advertising revenue, which makes us wonder if the plan is to sell surveillance to the highest bidder.

 
Inside the Atlas

Enough about our brave, bright, and continually monitored city of the future. What’s inside the box? The Atlas is powered by Aaeon’s NANOCOM-APL COM Express Type 10 Mini module equipped with a quad-core Intel Atom x7-E3950, the most powerful of the Apollo Lake SoCs. The module supplies 4GB to 8GB DDR4 and 32GB eMMC for the OS.

NanoCOM-APL

There’s also an option for adding an Apollo Lake based, dual-core Atom x7-E3930 via an Up Core Plus SBC for dedicated camera control and advanced edge analytics, thereby freeing up the NANOCOM-APL for other applications. It’s unclear if you can still control a camera linked to the Myriad X with the NANOCOM-APL alone.

The UP Core Plus based “Camera Module” includes an additional 32GB eMMC. The add-on is required for running the NVR application for smart parking, and it would appear to be required for an alternative use of the Atlas as an NVR/IoT gateway that aggregates data from multiple Atlas-powered streetlight edge nodes. The diagram farther above, however, hints that this IoT gateway might be another Aaeon embedded computer altogether.

UP Core Plus, front and back
(click image to enlarge)

The Atlas is equipped with an M.2-based 512GB SSD, which supports 15 hours of video at 3840 x 2160 resolution. There’s also a GbE port with PoE support, 2x USB 3.0 ports, and 8x USB, at least some of which are likely internal interfaces.

For displays, there’s a DDI port and 18/24-bit LVDS, which can optionally be swapped for higher-resolution eDP. Other I/O includes 2x UARTs plus I2C, SMBus, and 8-bit GPIO.

The Atlas supplies 3x mini-PCIe slots. Two are USB-only and support 3G/4G/LTE modules, such as a Telit LE910 4G/LTE add-on. A third mini-PCIe slot supports USB and PCIe x1. There’s also an Azure Wireless CM389MA module with WiFi, as well as built-in GPS.

The camera is an 80 x 22mm EverFocus EBN1840-A equipped with a 3.6mm fixed lens. The camera supports D-WDR and an IR control range of up to 30 meters “with 18 LEDs,” as well as “True Day/Night with automatic IR-cut.” Standard sensors on the Atlas include an ambient light sensor (ALS), a 3D G-sensor, a microphone, and an 8-megapixel progressive scan CMOS sensor.

The Atlas is powered with an 85-264V AC supply, or alternatively via PoE or a 12VDC input. Consumption is listed as under 40W. For streetlight control, there’s a 1000W lamp power on/off switch with a dimming function listed as “DALI and 0V to 10V.” The DALI interface supports many different types of lamps and lamp controllers. An energy meter monitors lamp power consumption.

The modest temperature range suggests that the Atlas is aimed more at Manila or Mumbai than Minsk or Milwaukee. Then again, Aaeon may simply be planning for global warming. The base level 0 to 60°C range drops to 0 to 50°C if you add the Up Core Plus and 0 to 45°C if you’re using environmental sensors.

The Atlas is IP66-protected against water and dust and offers 10% to 80%RH, non-condensing humidity resistance. Tamper-proofing and other security features are also available.

Analytics plug-ins include license plate detection, face detection, face recognition, vehicle color/type/maker/model classification, and perimeter intrusion prevention. Other software-related functions, which are supported with a web-based dashboard interface, include automatic lighting adjustment, lighting control priority, group lighting management, real-time malfunction alerts, mobile app support, and microclimatic air quality measurements for SO2, VOC, O3, NO2, CO, and PM2.5.

 
Further information

No pricing or availability information was provided for Aaeon’s Atlas. More information may be found in the announcement, as well as an Intel product brief (PDF) with a spec list and a second Intel product brief on smart lighting applications.