Crazyflie drones, developed by Bitcraze, have been a staple in research labs across the globe, including prestigious institutions like New York University and ETH Zurich. These drones are celebrated for their size, durability, accuracy, and most importantly, their open-source nature which allows for extensive customization and experimentation. Recognizing the educational potential of such advanced technology,

DroneBlocks has ingeniously adapted the Crazyflie drone for classroom use, creating a bridge between high-level research and foundational STEM education, incorporating them into their Autonomous Level II Drone Kit for K-12 schools. This move not only demystifies complex engineering and programming concepts but also makes cutting-edge technology accessible to students of all ages. With a web-based coding app, simulator, and comprehensive curriculum resources, DroneBlocks is making it easier than ever for educators to integrate drones into their STEM education programs. This initiative also aims to replace the now-discontinued DJI Tello drones, providing students of all ages with access to cutting-edge drone technology.

In this post, we’ll explore some groundbreaking projects from leading institutions that showcase the versatility and capabilities of these drones, particularly in the realm of autonomous flight.

Learning to Fly in Seconds: A Leap by ARPL at NYU

At the Agile Robotics and Perception Lab (ARPL) at New York University, researchers have embarked on an ambitious project titled “Learning to Fly in Seconds.” This project uses deep Reinforcement Learning (RL) to teach Crazyflie drones to control themselves with remarkable agility and precision, overcoming the notorious challenge of the simulation-to-reality gap. By introducing an asymmetric actor-critic-based architecture and a robust RL-based training paradigm, the team has managed to drastically reduce training times. This breakthrough enables the direct control of drones’ RPM (Revolutions Per Minute) with only 18 seconds of training on a consumer-grade laptop, showcasing an impressive trajectory tracking performance. The open-source nature of this project, along with the fast multirotor dynamics simulator, marks a significant step towards democratizing drone research and development.

GitHub | Watch on YouTube

Swarm Intelligence: Fully Onboard SLAM by D-ITET at ETH Zurich

The D-​ITET Center for Project-Based Learning at ETH Zurich is pushing the boundaries of swarm technology with its project on “Fully Onboard SLAM for Distributed Mapping with a Swarm of Nano-Drones.” Addressing key challenges in positioning, mapping, and communications, this project showcases how a swarm of nano-UAVs, each with a limited payload and onboard sensing, can generate coherent grid maps in real-time. With innovative use of depth sensors and low-power ARM Cortex-M microcontrollers, this research demonstrates scalable, efficient mapping of large areas, a crucial capability for surveillance, first-aid missions, and industrial automation. This project not only highlights the drones’ potential in collaborative tasks but also their capacity to operate without reliance on external infrastructure.

GitHub | Watch on YouTube

Autonomous Navigation on a Budget: “Catch Me If You Can” from ETH Zurich

In another pioneering project from ETH Zurich’s D-ITET Center, “Catch Me If You Can” explores autonomous navigation and obstacle avoidance with nano and micro UAVs. This project leverages a multi-zone Time of Flight (ToF) sensor, enabling drones to navigate autonomously with significantly lower computational demands than traditional visual-based solutions. The research showcases the potential of using simple, yet effective sensor technology to empower drones with reliable obstacle avoidance capabilities, essential for their widespread adoption in civilian and commercial applications.

Watch on YouTube | GitHub