Listen to this article

Chad Sweet is the Co-Founder and CEO of ModalAI, a San Diego based provider of AI-powered computing platforms for autonomous drones and robots that communicate via 4G and 5G cellular networks. ModalAI was spun out of wireless technology provider Qualcomm in 2018. Prior to founding ModalAI, Sweet was Senior Director of Engineering in Qualcomm’s Robotics R&D Group. He also held various engineering positions during his 20 years at Qualcomm, with 15 patents granted during that time. In February, Sweet was elected to the Board of Directors of The Dronecode Foundation, a vendor-neutral foundation for open source drone projects.

In April 2020, ModalAI released VOXL Flight, a development platform for GPS-denied autonomous drone localization and navigation, along with computer vision for AI-powered object recognition, depth sensing and obstacle avoidance. The SWaP-C (size, weight, power consumption, and cost) optimized VOXL Flight incorporates a Qualcomm Snapdragon SoC and PX4 flight controller on a single board, and can support up to four image sensors for simultaneous 4k video capture, streaming and machine vision processing. An LTE Add-on is available that plugs in to the VOXL’s board to board connector enabling command and control for operations beyond visual line of sight (BVLOS). ModalAI’s VOXL Flight was a 2021 RBR50 Robotics Innovation Award honoree.

Chad Sweet was recently interviewed by Oliver McClintock for My Dear Drone. That interview is reprinted below, with minor modifications. The original interview can be found HERE.

---

Oliver McClintock – Welcome Chad Sweet, tell us on yourself and your backstory?

Chad Sweet – After a long career in R&D engineering at Qualcomm in San Diego, in June 2018, Donald Hutson and I spun out the robotics R&D project into a separate company, ModalAI. I’m passionate about pushing the boundaries of the latest technologies, specifically autonomous vehicles. Robots excite me as the advances are so tangible. You make a change, and the results are immediately visible.

After receiving my Bachelor of Engineering in computer engineering at Vanderbilt University in 1998, I started at Qualcomm Inc. There I ultimately led global robotics R&D at Qualcomm Research working on multiple autonomous robotics product launches, including Sony Aibo, Hover Camera, Snapdragon Flight, and Ecovacs DJ35. Previously to that, I led efforts in computer vision and wireless communications. In total, I’m named on 15 patents.

Oliver McClintock – How did you get going in the robotic and drone markets?

Chad Sweet – At Qualcomm, I was an early leader in the computer vision efforts, some of which evolved into Vuforia and FastCV. I had the opportunity to work on a novel, neuromorphic approach to robotics, and I jumped on it. Autonomous robotics can be considered computer vision, so it was exciting to approach some older problems in a new environment.

That neuromorphic project transitioned to bringing Snapdragon to the robotics market. We quickly saw consumer drones as a growing market and addressed the market with Snapdragon Flight, the smallest high-performance computing platform for drones at the time.

Oliver McClintock – Tell us about your ModalAI company, what is it, and what motivated you to launch it?

Chad Sweet – ModalAI builds on the R&D we were doing at Qualcomm. We are making the smallest, most advanced flight control and communication systems for drones. ModalAI continues to advance the technology we developed at Qualcomm by commercializing more AI software and hardware. We have heavily invested in the open-source communities Dronecode and ROS. ModalAI’s goal is to make our platform as open as possible so that developers can integrate their novel applications on the platform as efficiently as possible.

ModalAI left Qualcomm with a SWaP-C optimized, market-ready product, now called VOXL. Built upon more than six years of R&D, VOXL leverages the smartphone environment to make a machine vision-based, highly-integrated, autonomous navigation platform. The platform provides autonomy and communications for indoor and outdoor ground and aerial robots. The integration provided by VOXL enables UAVs that are well under 300g in total weight to navigate autonomously.

Working with the Department of Defense’s Defense Innovation Unit (DIU), ModalAI has also developed a hand-held, private LTE network called ModalLink. ModalLink communicates with multiple UAS, robots, and smartphones using the DOD’s 1800MHz spectrum.

---

VOXL Flight fuses the flexibility and ease of use of an open-source flight controller with the sophistication of the state-of-the-art smartphone electronics ecosystem.

---

Oliver McClintock – Tell us on Flight platform you manufacture and design for autonomous UAV navigation?

Chad Sweet – VOXL Flight (Figure 1) is the first open-development platform for autonomous drone navigation that fuses a companion computer with a flight controller on one PCB for GPS-denied, autonomous PX4 drones, and robots.

It combines ModalAI’s VOXL and Flight Core technology to get equipment in the air quickly with less programming and installation time. The platform is also compliant with the National Defense Authorization Act of 2020 as ModalAI designs and manufactures all of its printed circuit board installations in Southern California.

VOXL Flight fuses the flexibility and ease of use of an open-source flight controller with the sophistication of the state-of-the-art smartphone electronics ecosystem. VOXL Flight has a Qualcomm Snapdragon multi-core processor on board for autonomous navigation, as well as an MCU to run the PX4 flight controller.

The tight coupling of the two reduces cabling, cost, size, and weight as well as other complexities, allowing easier integration into new drones. VOXL Flight supports several add-ons for communications such as 4G LTE for commercial enterprise applications and Microhard pDDL for military and industrial applications.

VOXL Flight is a platform others can use to build new applications on, but it also comes with many plug-and-play tools. The device ships with GPS denied navigation, video streaming, stereo depth maps, and support for QGroundControl.

Developer tools such as OpenCV, Docker, and TensorFlow Lite for deep learning AI are available at https://code.modalai.com.

---

---

Oliver McClintock – Who are most of your customers, or your most common projects types?

Chad Sweet – A broad range of commercial and military customers use ModalAI platforms to build different types of drones. Some customers want Unmanned Aerial Systems (UAS) for indoor navigation to conduct inspections. Others are looking for beyond visual line of sight command and control for commercial delivery applications or keeping personnel out of harm’s way.

Our specialty is working with customers to upgrade their existing platforms with VOXL to get their drones equipped with autonomous navigation and in the air quickly. ModalAI has involved with over a dozen customers to advance the platform developed at Qualcomm or commercialize the VOXL platform.

Since the year 2018, we’ve been cooperating with the DIU (Department of Defense’s Defense Innovation Unit) on architecture to Group 1 UAS (Unmanned Aircraft Systems). Group 1 UAS primarily weigh below 20 lbs, operate below 1200 above the ground level at speeds below 250 knots.

ModalAI designed a flight controller to the DIU depending on its VOXL system that uses AI (Artificial Intelligence) to operate autonomously, plan mission paths, and avoid obstacles – all connected through 4G LTE technology. Also, we developed a ground station with 4G/LTE, now called ModalLink, which allows troops to manage multiple UAS on the field. Such technologies enable lightweight, small, but advanced UAS with long-range, secure communications.

Equipment included with VOXL enables armed forces for mapping the ecosystem, other people, and drones while in a mission, and dynamically avoid obstructions. Unlike GPS-exclusive devices that can encounter signal drop off because of lost or weak satellite links, a VOXL-based drone gives armed troops with computer vision-centered navigation; thus, they can deploy their UAS wherever.

---

A typical remote control cannot reach more than 1 to 2 kilometers. Leveraging cellular technology enables communication as far as the mobile network will take you.

---

Oliver McClintock – When Uber Eats tested its initial food delivery successfully from a drone, VOXL by ModalAI was on-board.

Chad Sweet – Attached to the craft, VOXL utilized 4G connectivity to correspond between the Uber Elevate Cloud Services and drone, a proprietary, dynamic airspace management system which guides and tracks drone trips to take-off, hover, and independently land.

VOXL supplied command & control of the quadcopter and assisted in aircraft’s autonomous navigation and air traffic screening through ADS-B (Automatic Dependent Surveillance-Broadcast). Also, it mapped the Uber Eats drone’s progress, so that the pilot-in-command can see the aircraft over a digital map all the time.

Oliver McClintock – What type of technologies goes into making VOXL Flight pilot drones Beyond Visual Line of Sight?

Chad Sweet – The most important technology is cellular communication. The cellular network works as well in the air as it does on the ground. A typical remote control cannot reach more than 1 to 2 kilometers. Leveraging cellular technology enables communication as far as the mobile network will take you.

Cellular connectivity also enables better positioning with Real-Time Kinematic (RTK) GPS. The RTK correction information can be found on the internet. Since the drone is now connected, it can access that information.

Oliver McClintock – For a drone maker that is just beginning to integrate VOXL Flight to their items, how may they go around getting started?

Chad Sweet – The quickest path is to buy a ModalAI reference drone, such as the VOXL m500. Developers can use the VOXL m500 to start building right out of the box, yet it’s not intended to compete with a consumer drone. If you’re comfortable making a drone on your own, the VOXL Flight Deck is ready to mount, or the VOXL Flight printed circuit board (PCB) is available for purchase by itself.

Oliver McClintock – What are some of the most creative ways you have seen drones getting used?

Chad Sweet – I have always really liked the swarm drone work coming from the University of Pennsylvania. We worked with them a lot at Qualcomm. Their swarming drone systems without GPS are cool. Groups are using other techniques to use drones in concert as well, which is fun. I am always impressed with small, indoor, and autonomous.

Oliver McClintock – What are your views on the future of autonomous drones?

Chad Sweet – Our goal is to enable anyone to control multiple drones easily and to accomplish a compelling task, quickly, safely, and easily in any environment. Social distancing is a hot topic, and autonomous vehicles, in general, are a great way to limit unnecessary interactions. I expect to see the pace of deployment accelerate in the next few years as the benefits of saving time, lives, and money becomes more apparent.

Oliver McClintock – What are a few of the benefits of deploying drones in the defense and security industries?

Chad Sweet – The most impactful advantage of using drones for security and defense is getting surveillance data quickly without needing to send a person into harm’s way – remote eyes in the sky. In the long-term, drones will be used to perform real-time analysis using a variety of visual and environmental sensors in a more automated way. Drones will also augment communication systems, especially in a crisis.

---

The most exciting drone technologies enable smaller, smarter, and safer drones. During the coming years, drones will transition from a loud nuisance to a useful, secure tool for many tasks.

---

Oliver McClintock – What advances in drone systems you are excited about?

Chad Sweet – The most exciting drone technologies enable smaller, smarter, and safer drones. During the coming years, drones will transition from a loud nuisance to a useful, secure tool for many tasks. Those improvements will become especially meaningful for indoor applications.

Oliver McClintock – VOXL Flight is compliant with each federal acts of the United States. Amazing! How did you handle that?

Chad Sweet – Unfortunately, a lot of the manufacturing for this complicated technology has moved overseas. We have a TON of experience in the company to tackle this problem. Many of us worked at Qualcomm for more than two decades, so we were able to use that experience to build our technology in the U.S.

Oliver McClintock – When you are not working, where can we find out you?

Chad Sweet – My hobbies are running, skiing, and playing with my three boys. We are a very busy family, and there is always something going on!

Oliver McClintock – Is there anything else which we have not covered that you would love to share?

Chad Sweet – As I mentioned, my team and I are passionate about autonomous robots. For autonomous drones and robots to reach the tipping point for commercial use, the industry will need to conform and standardize making drones is simpler, faster, and more affordable.

ModalAI is looking to fill that gap. Our niche is designing full-stack technology and driving it all the way to the application layer. All the hardware is available online for purchase affordably on our website. And the team is very dedicated to supporting the technical documentation in an open and public forum.