Fast Track to Flight
Consumer and commercial drones pose a number of tricky design challenges. Technology vendors have made things somewhat easier over the past year, offering a variety of system-oriented platforms and tools—even including complete development kits.
The development of consumer and commercial drones continues to be a dynamic segment of the embedded systems industry. Faced with severe limits on size, weight and power, drone designers need to be careful with how they choose each and every electronic component. Meanwhile, huge opportunities abound for drone platforms that can pack in high levels of compute processing along with advanced cameras and sensor suites.
Fortunately, drone developers don’t have to start from scratch. A rich set of resources are available including board-level solutions, payload subsystems and development kits and even complete reference designs. Over the last 12 months, new solutions along those lines continue to roll out from a variety of vendors ranging from processor companies to drone vendors themselves.
SYSTEM MODULE SOLUTION
Exemplifying those trends, in October Intrinsyc announced its tiny Open-Q uSOM module. The new Open-Q 845 uSOM is a 50mm × 25mm mini-module is based on Qualcomm’s Snapdragon 845 SoC (Figure 1). It’s supported by a Mini-ITX form-factor Open-Q 845 μSOM development kit. The module is designed for advanced robotics, drones and embedded IoT devices requiring the latest on-device AI powers, says Intrinsyc. It runs the Android 9 Pie OS, with a promise to upgrade to the latest Android 10 by 2Q 2020. The module is also supported by a Yocto-based Linux image that is similarly based on Linux kernel 4.9.
Aside from the Open-Q 845 HDK for mobile phones released in 2018, the 8-core, 10nm-fabricated Snapdragon 845 SoC has appeared on the Robotics RB3 Platform from Qualcomm and Thundercomm, which is built around a DragonBoard 845c SBC that has yet to be released separately. More on the RB3 later in this article.
The Open-Q 845 uSOM module ships with 4GB or 6GB dual-channel LPDDR4x SDRAM at 1866MHz, as well as 32GB or 64GB UFS flash. There’s also a 2.4/5GHz 802.11a/b/g/n/ac wotj 2×2 MU-MIMO (Qualcomm WCN3990) with a 5GHz external PA and U.FL antenna connector. A Bluetooth 5.x radio is also included. Media interfaces include DisplayPort v1.4 with USB Type-C support for up to 4K60 and 2x 4-lane MIPI-DSI D-PHY 1.2 at up to 3840×2400 10-bit 60fps. Camera interfaces include 3x 4-lane MIPI-CSI and a separate 2-lane MIPI-CSI link.
The development kit for the Open-Q 845 μSOM is built around a 170m × 170mm carrier board. There’s also an optional smartphone sized touchscreen and 13-Mpixel camera. The Open-Q 845 μSOM Development Kit carrier runs on 12V/3A power via an included adapter and can also operate on a user-supplied Li-Ion battery.
The board provides a USB 3.1 Type-C port with DP and USB support and there are connectors for all the MIPI-DSI and -CSI interfaces mentioned above. Audio features include the WCD9340 codec, a 3.5mm audio combo jack and analog and digital audio I/O headers. The carrier has a microSD slot, a USB 3.1 host port and a PCIe Gen3 interface. There are also headers for UART, I2C, SPI and configurable GPIOs. Dual PCB antennas are also available.
Among the most compelling advances in commercial drone usage has been the integration of mesh-networks to enable drone communications. Rajant offers a technology solution along those lines. Using a combination of wireless network nodes that Rajant calls Breadcrumbs and its InstaMesh networking software, Rajant’s Kinetic Mesh networks employ any-node to any-node capabilities to continuously and instantaneously route data via the best available traffic path and frequency—for any number of nodes, all with extremely low overhead. Rajant BreadCrumbs can communicate with any Wi-Fi or Ethernet-connected device to deliver low-latency, high-throughput data, voice and video applications across the meshed, self-healing network.
In November, Rajant released its latest BreadCrumb product, the DX2. The DX2 is Rajant’s smallest and lightest BreadCrumb, forming a mesh network when used in conjunction with its LX5, ME4 and ES1 models, which operate using Rajant’s proprietary InstaMesh protocol (Figure 2). With one transceiver and two external antennas, DX2 is lightweight and has low power consumption depending on transceiver configuration. Encased in a magnesium enclosure, the DX2 weighs 123g making it well suited for lightweight autonomous vehicles, drones and small robots. This very low payload, combined with a pocket-size footprint, makes it a good solution for varying degrees of autonomy and mobility operations as well as high bandwidth communication and data transmission, according to Rajant.
The DX2 has integrated Wi-Fi access point service for compatibility with millions of commercial off-the-shelf (COTS) client devices, such as laptops, tablets, smartphones, IP cameras, sensors and other IP devices. Additionally, a hidden USB connector, to be used for GPS or Tactical Radio over IP (TRoIP), lies behind a rear black rubber plug.
In compatibility with all other Rajant nodes, the DX2 forms a wireless Kinetic Mesh network that maintains continuous connectivity unlike traditional break-before-make infrastructures, says Rajant. Like all other Rajant BreadCrumbs, the DX2 delivers low-latency, high-throughput, fail-proof connectivity for data, voice and video applications, including drone swarms. The DX2 is available in two models, the DX2-24 with 2.4 GHz and DX2-50 with 5.0 GHz.
Focusing on the high-bandwidth side of drone data transfer, Silvus Technologies provides communications solutions for high bandwidth video, C2, health and telemetry data. In September, Silvus announced a partnership with Silent Falcon UAS Technologies, manufacturer of the Silent Falcon, a solar electric, fixed wing, long endurance, long range drone (Figure 3). The Silent Falcon drone integrates Silvus’ advanced technology MIMO MANET Streamcaster communications systems in its drone systems including its new SF ATAK Field Observer Kit.
Introduced in 2019, the most recent models of the Streamcaster radios are Silvus’ Enhanced 4000 series. The new radios provide a user-customizable multilocation switch for loading presets and zeroizing crypto. They have improved connectors and tie-down points for weather caps and feature IP68 enclosures (submersible to 20m). Smart battery technology provides % monitoring. The units also feature FIPS140-2 Level 2 encryption and MANET Interference Avoidance (MAN-IA).
The SC4200E model in the Enhanced series is a 2×2 MIMO radio. It is well suited for use in portable and drone applications where size, weight, power or cost are key. The unit provides up to 4W of output power (up to 8W effective performance thanks to TX Beamforming). The SC4200E is available in three form factors to suit a variety of applications: Rugged “brick” (externally powered), rugged handheld (with twist-lock battery connector) and non-rugged OEM (for embedding in custom products and sub-systems).
Silent Falcon has previously used the Silvus MIMO MANET communications systems for a wide variety of drone long range commercial applications in oil and gas, pipeline, electric power transmission, mapping and surveying markets. It has also been successfully deployed in intelligence, surveillance and reconnaissance; search and rescue and long-range border patrol missions. It’s also been used in extreme environmental conditions while assisting the US Department of Interior in wildfire fighting operations.
Silent Falcon recently introduced its three radio SF TriAntenna Ground Control Station, powered by Silvus Streamcaster components. The system increases the reliability, connectivity and bandwidth of the Silent Falcon system. The comm system’s capabilities have been further enhanced by the addition of the SF ATAK Field Observer Kit, a small, portable kit that provides live streaming videos with map overlays on tablets and smartphones to operators on the ground who need this vital information in real time.
SNAPS AND MANIFOLD 2
Drones like DJI’s Phantom and Matrice models embed flight controllers that run a proprietary operating system. But, in 2015, the company announced a Manifold development computer for its Matrice 100 drone that runs Ubuntu on a Nvidia Tegra K1. In June 2019, DJI unveiled a more powerful Manifold 2 computer with a choice of Nvidia Jetson TX2 and Intel Core i7-8550U processors (Figure 4). Canonical followed up by announcing that, not only will Ubuntu 16.04 return as the pre-installed OS for the device, but that it will include support for Ubuntu snaps application packages.
Ubuntu snaps are containerized software packages that work interchangeably across embedded, desktop and cloud-based Ubuntu distributions. Found on embedded Linux devices ranging from LimeSDR boards to Orange Pi PCs, they offer built-in security, automated updates and transaction rollback support. They also come with an online marketplace for sharing and selling different snaps applications. The Manifold 2 will be the first drone system to offer snaps, which will enable its functionality to be “altered, updated and expanded over time,” according to Canonical. Snaps will make it easier to manage large fleets of drones, as well as develop vertical applications that can be shared and modified for other use cases.
Ubuntu offers DJI drone users support for Linux, Nvidia CUDA, OpenCV and ROS (Robot Operating System). The Ubuntu-driven Manifold 2 is well suited for the research and development of professional applications and can access flight data and perform intelligent control and data analysis. The Manifold 2 can be integrated on DJI enterprise drones including the Matrice 210 series and Matrice 600 series, as well as its separately available N3 Flight Controller and A3 Flight Controller. The computer can process complex image data onboard the drone and get results immediately and can program drones to fly autonomously while identifying objects and avoiding obstacles, says DJI.
The Manifold 2 can act either as a companion computer or as a control computer over the flight controller. The system can be integrated into the drone’s internal systems and sensors using DJI’s software development kit. The Manifold 2 offers users a choice of two processing platforms, both of which run Ubuntu 16.04 with snaps. The first is the “GPU Model” (Manifold2-G) with Nvidia’s Jetson TX2, which offers a more powerful, hexa-core update to the Manifold 1’s Nvidia Tegra K1.
DJI lists different applications for the two models. The GPU Model is said to be designed for AI, object recognition, motion analysis and image processing. The CPU Model is for autonomous flight, real-time data analysis, ground station connectivity and robotics. Both Manifold 2 versions have a -25°C to 45°C tolerant, 91mm × 61mm × 35mm enclosure, down from 110mm × 110mm × 26mm on the Manifold 1. Despite the smaller footprint, the new models are heavier than the under 200g original. The Jetson TX2-based GPU model weighs in at 230g while the Coffee Lake-based CPU Model is 205g.
COMPLETE DRONE DEV KIT
We’ve discussed several drone development kits in Circuit Cellar in recent years. These kinds of kits provide all the components needed to get a drone platform up and running. An example is the Smart Drone Development Platform from Aerotenna. The kit is equipped with microwave radar collision-avoidance sensors, a radar altimeter and an FPGA-based flight controller (Figure 5).
The kit’s OcPoC Zynq Mini Flight Controller is an FPGA-based flight controller capable of triple redundant GPS, compass and IMU. The unit is pre-loaded with the PX4 and Ardupilot flight control stacks. PX4 is the largest commercially deployed open source flight stack and supports contemporary airframe architectures including VTOL aircraft, multicopter and rover profile. The μLanding radar altimeter can perform in all weather conditions and challenging terrains and has maximum altitude range of 150m with and altitude accuracy of 2cm. Its update rate is 766Hz (every 1.31ms).
The development kit also includes three μSharp-Patch collision avoidance radar sensors. These sensors scan the front, left and right side of the vehicle, detecting and locating obstacles on the horizon quickly and reliably and a maximum range of 120m. A pre-assembled quadcopter, carbon fiber airframe is provided in the kit. It includes GPS/compass, foldable arms for ease of transport and modular component design for simple maintenance and repair. It can do flight times up to 50 minutes (16000mAh battery, no payload). Total weight with airframe, pre-assembled flight controller and sensors is 1.9kg. Maximum takeoff weight with battery is 4.0kg.
DEV KIT RUNS ROS
Among the drone development kits introduced in 2019 was the Robotics RB3 Platform co-developed by Qualcomm and Thundercomm (Figure 6). The platform includes an octa-core Snapdragon 845 via a new “DragonBoard 845c” 96Boards SBC and tracking cameras. While the platform appears to be marketed toward terrestrial robots, Qualcomm told us that it’s also suited for developing drones.
The RB3 platform integrates key capabilities such as high-performance heterogeneous computing, 4G/LTE connectivity including CBRS support for private LTE networks, a Qualcomm AI Engine for on-device machine learning and computer vision, hi-fidelity sensor processing for perception, odometry for localization, mapping and navigation, advanced security and Wi-Fi connectivity. Support is also planned for 5G connectivity.
The platform currently supports Linux and Robot Operating System (ROS), while also including support for the Qualcomm Neural Processing software development kit (SDK) for advanced on-device AI, the Qualcomm Computer Vision Suite, the Qualcomm Hexagon DSP SDK and Amazon’s AWS RoboMaker, with plans for Ubuntu Linux support.
The platform’s hardware development kit contains the new purpose-built robotics-focused DragonBoard 845c development board, based on the Qualcomm SDA/SDM845 SoC and compliant with the 96Boards open hardware specification to support a broad range of mezzanine-board expansions. Optional elements for the kit include a connectivity board; an image camera for superb hi-res photo, 4K video capture and AI-assisted detection and recognition of people and objects; a tracking camera for path planning and obstacle avoidance using visual simultaneous localization and mapping (vSLAM); a stereo camera for navigation; and a time-of-flight camera for people, gesture and object detection even in low light conditions.
KIT FOR DRONE CHALLENGE
There’s been a history of processor vendors providing drone development kits—Intel and Qualcomm, for example. NXP Semiconductors for its part, has put a twist on this trend by making a drone development kit part of an annual drone development contest. Called the HoverGames Challenges, participants use NXP’s HoverGames drone development kit. The hardware and software of the developer kit is open, flexible and modular and includes professional, automotive and industrial-grade components enabled by the PX4 flight stack.
The HoverGames KIT-HGDRONEK66 kit (Figure 7) provides the mechanical and other components needed to evaluate the RDDRONE-FMUK66 flight management unit and adds BLDC motor control capabilities and a mechanical platform, on which it can be mounted. This developer kit may be used as part of and contains the components needed for the HoverGames coding challenges. NXP points out that this is a professional developer kit, not a complete functional system and includes no software. The flight management unit (FMU) is supported by the business-friendly open source PX4 flight stack. In addition, a separate suitable hobby-type LiPo battery and country-specific telemetry radio will be required.
When assembled, the frame has appropriate the additional space necessary to mount other components such as an adapter for Rapid IoT, NXP Freedom boards, or a companion computer such as i.MX 8M Mini to be used as a vision processor running Linux and ROS. The HoverGames drone and rover development platform is very flexible, fully open for development of robotics, control algorithms, security networking and communications protocols and can include another add-on component, companion computer, software or associated solutions.
Today’s quadcopter style consumer and commercial drones couldn’t exist without today’s high levels of chip integration. As developers push for more autonomous operations and AI aboard drones, they’ll continue to look toward SoC-based solutions to offer improved functionally without added size and weight. Fortunately, technologies and solutions such as those covered in this article can help drone system developers to get to market—and to flight—faster.
Aerotenna | www.aerotenna.com
Intrinsyc Technologies | www.intrinsyc.com
Nvidia | www.nvidia.com
NXP Semiconductors | www.nxp.com
Qualcomm | www.qualcomm.com
Rajant | www.rajant.com
Silvus Technologies | www.silvustechnologies.com
PUBLISHED IN CIRCUIT CELLAR MAGAZINE • JANUARY 2020 #354 – Get a PDF of the issue