I recently met up with Vasu Madabushi, Director of Marketing and Business Development at Inforce Computing (Silicon Valley). I found out to my surprise that InForce is one of the few Qualcomm partners worldwide that designs and manufactures embedded computing platforms. They design and manufacture Commercial Off-The-Shelf (COTS), product-ready system on modules (SOMs) and single-board computers (SBCs) that feature ARM® ISA-compliant Qualcomm® Snapdragon™ processors.
Snapdragon processors have been used in mobile devices and high-end smartphones, so it was delightful to see the same power-efficient high-end processors being deployed in embedded designs. What I wasn’t fully aware of was some of the cool devices these processors have gone into lately. Inforce’s solutions have found their way into augmented reality (AR) head-mounted displays, portable ultrasound imaging systems, devices that require high-end 4K Ultra HD and 1080p hardware-assisted video capture/playback and image signal processing, enterprise videoconferencing, and of course robots and drones that run heavy-duty computer vision algorithms.
What sparked my interest further during the conversation is the knowledge that Vasu shared on drones and robotics. Certainly this is a hot topic with companies like Google buying up a number of robotics firms and working on autonomous driving technology. Add in Amazon with its research on drones and the recent Qualcomm Snapdragon “flight” reference design. All of this resonated with me as I spent a lot of time working in the automotive industry on Advanced Driver Assistance Systems (ADAS), which are the building blocks to autonomous driving, and I also featured the drone maker, YUNEEC, in my pre-con at Sensor Expo 2015. But I digress…..
Vasu said there is a whole open source community for robotics and drone development. For example, the Robot Operating System (ROS) from Open Source Robotics Foundation (OSRF) provides a set of software libraries, drivers, developer tools, and algorithms for building robotics applications, so one doesn’t have to start from scratch. OSRF has ported the ROS to the Inforce 6410Plus SBC, making it easier for others to adopt a Snapdragon processor-based embedded system [Read ROS Support for Qualcomm® Snapdragon™ - Bringing ARM into Robotics for more information].
The Linux Foundation’s DroneCode aims to do a similar thing for enabling open-source projects that include flight control, mission planning, and advanced functionality such as computer vision, navigation and obstacle avoidance, and real-time flight planning. Open-source hardware projects include an autopilot function.
There are lots of applications potentially running on a drone [Read “Commercial Drones: Where’re the Viable Use-cases and Core Technologies to Differentiate in a Crowded Marketplace?"]. For example, drones could constantly capture high-resolution images and data from a wide range of on-board sensors, which necessitates quite a bit of real-time and heavy-duty processing. Furthermore, running network stacks for connecting with the cloud also puts a strain on power consumption and payload capacity. Of course, computing capability is at a premium for all of these CPU intensive applications, so it is no wonder many embedded designers are looking to get high performing small form-factor compute platforms--such as the Snapdragon with its added power-efficient heterogeneous hardware engines such as GPUs, DSPs and ISPs--integrated in the SoC. For example, the Snapdragon 600 delivers close to an astonishing 23,000 DMIPS (integer performance) and the even more advanced Snapdragon 805 get you up to 38,000 DMIPS. This kind of processing power available at mobile power efficiencies makes an ideal fit for high-end embedded applications such as those demanded by Drones and UAVs.
If you are interested in a Qualcomm Snapdragon powered SOM you need to check out Inforce Computing.