American dining habits are changing, as busy schedules and varying tastes increase demand for ordered food. New technologies enable ordering by smartphone apps and delivery by robots or aerial drones. High-density urban areas pose a challenge for drone delivery, so Uber Eats used ModalAI’s VOXL platform in its tests here this summer.
The average U.S. customer of Uber Eats spends more than $220 on orders annually, said ModalAI. The market for robotic food delivery could triple to $34 million by 2024, predicts Markets and Markets. In addition to drone deliveries, developers are working on mobile robots and remotely controlled and autonomous vehicles.
ModalAI takes off from Qualcomm efforts
ModalAI CEO Chad Sweet led Qualcomm Technologies Inc.’s drone research and development before formally spinning out the startup last year.
“Qualcomm has been a tremendous partner and has been generous with time and support since then,” he told The Robot Report. “We’ve been productizing state-of-the-art technology for drones and robotics while Qualcomm focuses on its Snapdragon chips.”
“We’ve taken a developer-centric approach, to make our compact, lightweight, high-performing VOXL as easy to use as other platforms in this space,” said Sweet. “We’re building a ton of developer tools and are part of the drone code community.”
“Another big advancement is that we’re really fusing the cellular technology, baking that in as a first-class feature,” he added. “Since the drone is connected to the Internet, it can talk to other things, which played into the Uber effort. A drone looks like a car in Uber’s app.”
The drone delivery test was part of the U.S. Federal Aviation Administration’s Unmanned Aerial Systems Integrated Pilot Program (IPP). ModalAI joined the IPP program in 2018 in coordination with Qualcomm.
VOXL ready to support applications
Unlike flight systems that have difficulty with weak or lost satellite connections, VOXL uses smartphone networks rather than the Global Positioning System (GPS). The company said this allows for an integrated, machine vision-based, autonomous navigation system for drones and robots, both indoors and outdoors.
“Integrating with different cameras is harder than you might think,” Sweet said. “Thanks to more than 100 years of cumulative Qualcomm experience on our staff, we can leverage the smartphone ecosystem.”
“This package includes computer vision and processing capable of running SLAM [simultaneous localization and mapping], a flight controller, and connectivity for other peripherals,” he added. “The flight controller is typically on different chips, but we can run it within Snapdragon, isolated from the CPU.”
“VOXL supports four cameras over MIPI [a widely used interface], and we’ve built a USB hub into the platform,” explained Sweet. “We’re doing the hard stuff, so that integrators can take the last step and enable the platform for lots of different use cases.”
Uber Eats and BVLOS
In the San Diego test, the drone used VOXL to communicate over 4G to Uber Elevate Cloud Services, a proprietary system that tracks and guides drone flights. The system provided command and control of the drone and assisted with autonomous navigation and air traffic monitoring via an Automatic Dependent Surveillance-Broadcast (ADS-B) receiver.
VOXL also provided a digital map so that the remote pilot could see where the drone was at all times, even when it was beyond visual line of sight (BVLOS).
What did ModalAI learn from the Uber Eats test? “The technology is there — we feel really good about that,” said Sweet. “It’s just like the FAA’s approach to BVLOS. There needs to be a lot of controlled data collection and experiments before the public is comfortable. The more opportunities there are to prove that the technology is safe, the better.”
VOXL lifts off with cellular connections
While ModalAI is not currently working on systems to take advantage of promised 5G networks, “we’re absolutely watching it,” said Sweet. “I would expect everything we’re doing in 4G to be available and more integrated.”
“With our product, you can plug in and fly drones that are cellular-connected,” he said. “That’s a differentiator from other traditional drone platforms. Ours is also more extensible to ROS [the Robot Operating System] and other software. You can do obstacle avoidance and mapping all in a small, lightweight package.”
ModalAI has other customers, including the Defense Innovation Unit of the U.S. Department of Defense (DOD).
“When the DOD grounded DJI drones because of security concerns, it thought it would be easier to find replacements than it was,” Sweet said. “Just as with the drone industry’s pivot from consumer to commercial, the priority shifts from trying to build something for $300 to making sure it brings value to the enterprise.”
“There are a lot of products on the market that are similar, such as the NVIDIA Jetson, but folks have neglected these platforms after the consumer market stagnated,” he said. “Jetson weighs 150g [5.2 oz.], while our product weighs 20g [0.7 oz.] with cameras and communications built in. We’re the only company fully integrating cellular capabilities into the drone and robotics space.”
Source: The Robot Report