Agilicious Is an Open Source, NVIDIA Jetson-Powered Agile Quadcopter Drone Platform
Offering vision-based low-latency navigation at up to 5g and 70km/h, this drone platform can fly rings around the competition.
A team of researchers at the University of Zurich has announced plans to release Agilicious, an open source platform for highly-agile quadcopter drones — navigating entirely by vision, processed through an NVIDIA Jetson TX2 system-on-module.
"Agilicious is a co-designed hardware and software framework tailored to autonomous, agile quadrotor flight," its creators explain of the project, "which has been developed and used since 2016 at the Robotics and Perception Group (RPG) of the University of Zurich. It is completely open source and open-hardware and supports both model-based and neural-network-based controllers."
The platform promises much — including high thrust-to-weight and torque-to-inertia ratio, built-in vision sensors, GPU acceleration for real-time computer vision perception and neural network inference, a real-time flight controller, and a quick-start simulation system.
The drone itself is based on a quartet of 5" propellers attached to brushless motors under the control of a HobbyWing XRotor electronic speed control (ESC) board and BrainFPV Radix flight controller running the team's in-house agiNuttx software stack, all assembled on an Armattan Chameleon 6 frame. Any one of a range of cameras, open source or proprietary, can be used to feed visual data to the NVIDIA Jetson TX2 system-on-module — hosted on a ConnectTech Quasar carrier board.
The Robot Operating System (ROS)-compatible software stack, meanwhile, is designed for minimal latency to offer high agility — and, in testing, certainly seems to deliver. "Our demonstrators include trajectory tracking at up to 5g and 70 kilometers per hour in a motion capture system," the team writes, "and vision-based acrobatic flight and obstacle avoidance in both structured and unstructured environments using solely onboard perception."
The best part of the Agilicious platform, though, is that both the software and hardware stacks are fully open source — or, rather, they will be. The team has confirmed plans to release all the design files and source code under the reciprocal GNU General Public License 3; at the time of writing, however, the project's GitHub repository was empty bar a readme, license file, and references.
To keep you occupied while you're waiting for the files to be uploaded to GitHub, the team's paper describing Agilicious is available under open-access terms in the journal Science Robotics.