This repository contains the ROS implementation of the Composite CBF Safety Filter for Collision Avoidance for Multirotors.
It computes the optimally minimal modifications to an acceleration setpoint that prevent collisions based on velocity estimates and depth measurements.
See our paper video on YouTube!
The filter operates using real-time onboard range measurements and velocity estimation. Specifically, the filter operates on a downsampled point cloud from the onboard sensor mounted on the robot, including, e.g., Intel RealSense D455 and alike, small-scale Time-of-Flight sensors, or LiDARs. We release a point cloud selector node for generic depth / range sensors on this dedicated repo.
The node implements the unconstrained-FoV case. It relies on computing an analytical solution to the underlying CBF-QP, which is computationally efficient.
We also release an embedded implementation for PX4.
- ROS 2 (Tested on Humble)
- mavros-msgs
- Eigen3
- Clone the repo in the
srcfolder of your workspace and install it using Colcon. - Install the point cloud selector node.
The yaml config files are loaded as ROS2 parameters in the node.
The CBF tuning is presented in our paper, Section V.E.
We report here the table detailing the config files:
| Parameter | Range | Effect when increasing |
|---|---|---|
epsilon |
>0 | Larger avoidance radius |
kappa |
[10, 100] | Less smooth approximation |
gamma |
[10, 100] | Reacts to farther obstacles |
alpha |
[1, 3] | Increased filter sensitivity |
pole_0 |
[-3, -1] | Damped response |
lp_gain_out |
[0, 1] | Lesser acc. smoothing |
max_acc_xy |
>0 | Larger maximum output xy acc. |
max_acc_z |
>0 | Larger maximum output z acc. |
obs_to |
>0 | Longer timeout for old input |
cmd_to |
>0 | Longer timeout for old ref. |
If you use this work in your research, please cite the following publication:
@INPROCEEDINGS{harms2025safe,
AUTHOR={Marvin Harms and Martin Jacquet and Kostas Alexis},
TITLE={Safe Quadrotor Navigation using Composite Control Barrier Functions},
BOOKTITLE={2025 IEEE International Conference on Robotics and Automation (ICRA)},
YEAR={2025},
URL={https://arxiv.org/abs/2502.04101},
}Or, if you use our embedded implementation, please cite:
@INPROCEEDINGS{misyats2025embedded,
AUTHOR={Misyats, Nazar and Harms, Marvin and Nissov, Morten and Jacquet, Martin and Alexis, Kostas},
TITLE={Embedded Safe Reactive Navigation for Multirotors Systems using Control Barrier Functions},
BOOKTITLE={2025 International Conference on Unmanned Aircraft Systems (ICUAS)},
pages={697--704},
YEAR={2025},
URL={https://arxiv.org/abs/2504.15850},
}This work was supported by the European Commission Horizon Europe grants DIGIFOREST (EC 101070405) and SPEAR (EC 101119774).