Skip to main content

VOXL Flight by Silver member ModalAI, an open-development platform for autonomous drone navigation

By June 15, 2020Announcements
VOXL Flight by Silver member ModalAI

The San Diego based ModalAI spun out of Qualcomm® in 2018 and is leveraging the five years of prior research and development in the drone industry. 

The team at ModalAI is putting all that experience to work on VOXL Flight. They developed a plug-and-play open-development platform for GPS-denied autonomous drones and robots, running PX4, capable of autonomous drone navigation.

ModalAI offers a combination of companion computer and flight controller for safe indoor flying. VOXL Flight enables a wide range of applications, from indoor navigation to large scale, 4G-connected BVLOS.

Learn more about VOXL Flight on their Press Release below

VOXL Flight Combines Companion Computer and PX4 Flight Controller on One PCB For Safe Indoor Flying

By Chad Sweet, CEO ModalAI

Imagine the computing power of a Qualcomm Snapdragon SoC and the flexibility and ease-of-use of a dedicated PX4 flight controller on a single PCB. That’s the new VOXL Flight from ModalAI. 

VOXL Flight uses Visual Inertial Odometry (VIO) to fly safely indoors. With computer vision stability control, the drone flies more precisely, which makes it easier for the pilot to command. To optimize its VOXL platform for VIO, ModalAI employs camera and IMU frame time synchronization and exposure control algorithms. The 6 degrees of freedom position is computed and then passed directly to PX4 using the voxl-vision-px4 library and MAVLink.

Fundamentally, VIO combines a computer vision camera with measurements from an Inertial Measurement Unit (IMU). The magic is that it computes real-world coordinates in millimeters, from combining 3D features tracked in pixels and acceleration measured in meters/second2. The VIO algorithm publishes a covariance matrix that describes the certainty of its velocity and position estimates. 

VOXL Flight also supports Collision Prevention. In this case, depth maps are computed using stereo cameras, an obstacle map is computed, and the mapping data is passed to PX4 using MAVLink. The processing and communication are all done onboard VOXL Flight, and aside from tracking and stereo sensors, no additional hardware is needed.

The VOXL m500 development drone has VIO and Collision Prevention ready to fly and is preconfigured with camera mountings and wiring to mount on a UAV directly. The pre-configuration helps you get your drone in the air quickly with less programming and installation time.

VOXL Flight integrates the following technologies:

  • Artificial Intelligence (AI) for movement, such as Simultaneous Localization and Mapping (SLAM), to orient your drone to its ever-changing surroundings and plan mission paths
  • GPS Denied Flight capabilities to keep your drones safely in flight and navigating indoors and challenging outdoor environments
  • AI perception, using GPU accelerated TensorFlow Lite, facilitates obstacle avoidance with deep learning object recognition and integrated depth sensing 
  • Support for up to 4 image sensors for simultaneous 4k video capture, streaming and computer vision processing
  • 4G cellular or Microhard pDDL command, control and video via a plug-in add-on.

Click here to learn more.

Ramón Roche

Author Ramón Roche

Ramon is Dronecode's General Manager, you can read more about him on LinkedIn

More posts by Ramón Roche