This project contains an implementation of the L1 Adaptive Control for Crazyflie.
If you're new to flying with Crazyflie, follow their official documentation to assemble the drone, download and install the client (cflib), build and flash the firmware, and configure radio permissions. You should be able to connect to the Crazyflie remotely using the Crazyradio before you move on. For more information see Bitcraze's 'getting started' page here
- Hardware setup
We strongly encourage you to use some additional sensors to help the Crazyflie with localization. The simplest way to do this is with an Optical Flow deck (linked here). If you are using some sort of external localization tool such as Vicon, Optitrack, or any native Crazyswarm solutions (like LPS or Lighthouse) that's fine too. Without this additional sensing capability the Crazyflie will struggle to generate an accurate state estimate, and some controllers may fail entirely.
- Install dependencies
Our biggest dependencies are with the Crazyflie ecosystem. Start by installing the client and a toolchain to build the firmware with. Links can be found below
- client -- work up until the 'Extra' section, then return here
- firmware -- follow the directions to install an ARM toolchain, then return here. we'll cover the rest of the content on this page later.
From here, we add a select few dependencies of our own.
pip install pyyaml
- Clone this repository
git clone --recursive https://github.com/cfc-ray/L1-Crazyflie.git
- Build the firmware
The L1 adaptive controller has been implemented using the app layer in the Crazyflie firmware. This means there are no modifications needed to the main firmware - our code exists as an add-on. Follow the steps below to build the L1 adaptive controller alongside the main firmware.
cd L1-Crazyflie/controller_L1/
make -j 12
- Flash the firmware to your Crazyflie
You should now be in the L1 controller directory controller_L1. Power on your Crazyflie and connected a Crazyradio to your computer. Then run the following command.
cfloader flash build/cf2.bin stm32-fw -w <radio address>
where <radio address> should be replaced with the radio address of your Crazyflie. For example, if your Crazyflie is operating using radio number 01 and channel number E7, you will enter the above command as cfloader flash build/cf2.bin stm32-fw -w radio://0/01/2M/E7E7E7E7E7
At this point you should be ready to go to start flying a Crazyflie with the L1 Adaptive Control.
There is one configuration file: config.yaml, located in the flight_commands/ directory. Edit the uri in there to match the uri of your Crazyflie.
All that's needed is to run one of the flight scripts in the flight_commands/ directory. We provide a simple hover script that allows you to compare the stock Mellinger controller to that same Mellinger controller augmented with the L1 adaptive control. To use it, run the following command.
cd flight_commands
python3 hover.py
Follow the directions printed out in the terminal to turn the L1 adaptive control on and off. The image below depicts an example of the correct terminal output if everything is set up and running correctly.
It is very straightforward to integrate L1-Crazyflie into a current project. The L1 controller is implemented as an out-of-tree controller in the firmware, so it can be used with any flight control scripts. Simply change the active low level controller by switching the stabilizer/controller parameter to 5, and the L1 Adaptive Control will be activated.
Please read the license attached to this repository
For more information about the L1 Adaptive Control check out the following publications:
- Z. Wu, S. Cheng, P. Zhao, A. Gahlawat, K. Ackerman, A. Lakshmanan, C. Yang, J. Yu, and N. Hovakimyan, "
$\mathcal{L}_1$ Quad:$\mathcal{L}_1$ Adaptive Augmentation of Geometric Control for Agile Quadrotors with Performance Guarantees," arXiv preprint arXiv:2302.07208 (2023). (link) - N. Hovakimyan and C. Cao,
$\mathcal{L}_1$ Adaptive Control Theory: Guaranteed Robustness with Fast Adaptation. Philadelphia, PA, USA: SIAM, 2010. (link)
For more information about the Crazyflie platform check out the official Bitcraze documentation: