The benefits of autonomous driving are often touted and well known: they can cut carbon emissions, reduce accidents and create more efficient transportation economies in cities around the world. However, to truly scale the benefits of autonomous driving to citizens globally and not just the privileged few, we need an innovative approach towards lowering the cost of autonomous vehicles while remaining uncompromising on safety.
As a fellow autonomous driving researcher and practitioner, my personal mission is enable autonomous vehicles for citizens globally who potentially run the risk of being excluded from the pending transportation revolution. We believe in enabling researchers and practitioners like yourself to easily create your own autonomous vehicles to serve different needs, such as farming machines, delivery vehicles, etc.
DragonFly Car Modular Design
As shown in the Figure below, we believe in modular design, such that with a few simple modules we enable you to easily build an autonomous vehicle. The basic modules include a GNSS receiver for centimeter-level localization; a computer vision module for localization (through visual inertial odometry) and for visual perception; a planning and control module for making real-time motion planning; radars for mid-range obstacle detection; sonars for near-range obstacle detection; a cargo-carrying or passenger chassis; and a CAN bus to integrate everything.
DragonFly Computer Vision Module for localization and active visual perception
First, I would like to introduce DragonFly Computer Vision Module, our core sensor module, designed for visual perception and real-time localization tasks for autonomous vehicles. Specifically, DragonFly Computer Vision Module consists of:
1. Four-way hardware-synced global shutter images (720P) and a built-in Inertial Measurement Units (IMU)
2. Interface for acquiring RTK GPS data
3. Programmable NVIDIA Jetson SoM
4. Advanced VIO and Visual Perception packages
DragonFly 77 GHz millimeter Wave Radar for mid-range obstacle detection
As shown in the demo, for mid-range obstacle detection, we can apply 77 GHz millimeter wave radar such that the planning and control module can make decisions when obstacles are detected.
DragonFly Sonar for near-range obstacle detection
DragonFly sonars cover near-range obstacles and act as the very last line of defense, once sonars detect an obstacle, it directly signals the chassis to stop to minimize risks of an accident.
DragonFly GNSS for centimeter-level localization
Real-Time Kinematic (RTK) GNSS can achieve centimeter-level accuracy and thus it is suitable for autonomous driving applications. We provide reliable GNSS receiver modules (as well as base station modules) to help you achieve reliable real-time localization. At times when GNSS signal is blocked, we can fuse GNSS and DragonFly Visual Inertial Odometry to improve localization reliability and robustness.
DragonFly Chassis
We provide different chassis to fit your applications needs, we have developed a passenger carrying pod, an exploration tank, and a cargo carrying chassis. Talk to us about your application needs and we can work together to build the best solutions for you.
Build Your Own Autonomous Driving Applications
It will be a privilege for us to have you work with us to enable autonomous driving in different sectors. It could be a farming machine, an industrial vehicle, an exploration vehicle, or any other wild ideas you have in mind. Together let us democratize autonomous driving and robotic technologies!
Shaoshan Liu