Companion computers are now becoming an essential part of the drone ecosystem. The reason for their increased usage is quite clear. They allow users to develop complex applications smoothly and provide sublime control over the flight controller. Before companion computers, developers had no option but to toy around with embedded code of the flight controllers to develop drone applications. Now with the advent of companion computers, drone developers can build advanced drone applications with onboard intelligence, seamless connectivity to cloud over 4G/LTE, integrate custom payloads and build & connect custom Web/Mobile Interfaces.
Each of the below mentioned Companion Computers, have their own unique strengths and weaknesses. You can choose the one that best suits your drone applications.
NVIDIA TX1 – This is one of the most powerful companion computers, capable of delivering the performance and efficiency needed for advanced computing needs. Powered by 64-bit quad-core ARM cortex- A57 CPU and a 256-core Maxwell GPU, this board harnesses the power of deep learning and computer vision to deliver smart, autonomous learning machines.
Using this board to power your drones, can ensue some amazing drone applications. One can implement deep learning models to do feature detection, object recognition, perform real-time SLAM and even improve 3D mapping capabilities while using depth cameras. The NVIDIA TX1 delivers it all, flawlessly.
NVIDIA TX2 – Jetson TX2 is an improved version of the Jetson platform. It offers twice the efficiency resulting in twice the performance of Jetson TX1 AI applications. It features an integrated 256-core NVIDIA Pascal GPU, a hex-core ARMv8 64-bit CPU complex, and 8GB of LPDDR4 memory with a 128-bit interface. It also boasts a dual-core NVIDIA Denver 2 alongside a quad core ARM cortex-A57.
Jetson TX2 is also drop-in compatible with Jetson TX1 and provides an easy upgrade opportunity for products designed with Jetson TX1. (as mentioned in Nvidia dev blog here)
Intel Edison is a computer-on-module offered by Intel. Its small and compact size makes it ideal companion computer for drone application development. It comes with a dual-core Intel Atom processor at 500MHz, 1GB DDR3 RAM, 4 GB eMMC flash and is compatible with various programming languages.
Intel Edison has integrated Wi-Fi and Bluetooth, and can host an onboard server, enabling applications that require seamless wireless/cloud connectivity. The new Cube Autopilot aka Pixhawk2.1 now comes with integrated Intel Edison module. Intel recently announced that they are now discontinuing the Intel Edison board.
Intel® Aero is specially designed for UAVs. It is powered by a quad-core Intel® Atom processor and comes with direct support for Intel® Realsense technology, making it capable of performing object identification & avoidance and handling payload computation. It also hosts a Pre-programmed flight controller with Dronecode PX4 software (see here). The board is a size of standard playing card that combines compute, storage, connectivity and flexible I/O.
Intel® Joule is the latest companion computer offered by Intel. It is a high performance x86 board in tiny package. Given its compact shape and size, it is used extensively for computer vision and other applications that require high-end edge computing. It comes with quad-core 64-bit Intel Atom processor and has Intel® dual band wireless AC 8260 Wi-Fi/Bluetooth. Users have the options to choose from two Joule kits: the 550x which has 1.5GHz quad-core Atom processor, 3GB of RAM and 8GB of eMMC storage; the other one is 570x with a 1.7GHz quad-core Atom T5700 CPU(burst speed of 2.5GHz) 4GB of RAM and 16GB of eMMC storage. It comes with Intel® Realsense support. Intel recently announced that they are now discontinuing the Intel Joule board
Odroid XU4 by HardKernel is a smart computing device providing powerful and energy efficient hardware. It hosts a Samsung Exynos5422 cortex octa-core CPUs and built-in Mali-450 GPU. This enables implementation of complex onboard algorithms and computer vision applications. Along with uSD card, the eMMC 5.0 option provides faster read/write speeds. Multiple USB 3.0 , USB 2.0 and a Gigabit Ethernet port which makes payload integration and connectivity easy.
Raspberry Pi3 is a third generation Raspberry Pi. It comes with 1.2GHz 64-bit quad-core ARMv8 CPU, wireless LAN and Bluetooth support. This SBC is widely used by students, researchers and hobbyists to build different types of applications. It works smoothly when it comes to streaming live videos over wifi/cloud. Thus, a lot of computer vision based drone applications can be built using RPi3.
Qualcomm® Snapdragon Flight™ Kit (Developer’s Edition) is small (58x40mm) but offers a lot of CPU power and two on-board cameras. It contains 4 ‘Krait’ ARM cores which run Linux (Ubuntu 14.04 Trusty, by default), and 3 ‘Hexagon’ DSP cores which run the QURT RTOS. In addition, it includes Wi-Fi, Bluetooth connectivity, automotive-grade GPS and many more features.(as mentioned in ardupilot wiki)
FlytBase provides drone SDK and APIs for all these companion computers. It helps you setup your companion computer real quick and provides you with right set of developer tools(Mobile SDK, Web SDK, Drone Simulator and more) to build business-ready drone applications. You can refer to the detailed documentation and follow step-by-step guide to get started.
Visit flytbase.com to download FlytOS for free and accelerate your drone development.
By Prasad Tambekar