Code for Submission: "Lightweight Tracking Control for Computationally Constrained Aerial Systems with Newton-Raphson Flow"

Video footage for certain flights may be seen here

Paper: "Lightweight Tracking Control for Computationally Constrained Aerial Systems with the Newton-Raphson Method", preprint available
Authors: Evanns Morales-Cuadrado, Luke Baird, Yorai Wardi, Samuel Coogan
Venue: IEEE Transactions on Control Systems Technology (TCST), 2026

Environment Setup

1. Users of this code must use the /src/ folder here in a ROS2 workspace and build
2. Users may also use the environment.yml file to set up a conda environment with the necessary packages to run the code below
3. Docker (in development): A Docker-based workspace is available at ws_px4_work. Note that this is still under active development, but it aims to simplify the full PX4/ROS2 environment setup

Cloning This Repository

This repository uses git submodules for the quadrotor packages. Clone with:

git clone --recurse-submodules git@github.com:evannsmc/MoralesCuadrado_Baird_TCST2025.git

If you already cloned without --recurse-submodules, initialize the submodules with:

git submodule update --init --recursive

Motion Capture Packages (Optional)

If you are using a motion capture system for hardware flights, you may need the following packages cloned into your ROS2 workspace src/ directory:

# Vicon
git clone git@github.com:evannsmc/vicon4px4.git

# OptiTrack
git clone git@github.com:evannsmc/optitrack4px4.git

# PX4 messages (minimal branch)
git clone -b v1.16_minimal_msgs git@github.com:evannsmc/px4_msgs.git

# Mocap messages and relay nodes
git clone git@github.com:evannsmc/mocap_msgs.git
git clone git@github.com:evannsmc/mocap_px4_relays.git

Then build from the workspace root:

cd ..  # back to workspace root
colcon build --symlink-install

---

Quadrotor Code Instructions

The quadrotor packages (quad_newton_raphson_flow and quad_mpc) are included as git submodules pointing to their standalone repositories:

• NR Flow: newton_raphson_px4
• NMPC: nmpc_acados_px4

Preliminary

1. Follow the instructions (here) to set up the PX4 Autopilot Stack, ROS2, Micro XRCE-DDS Agent & Client, and build a ROS2 workspace with the necessary px4 communication repos
2. In the same workspace with the communication folders as above, go to the /src/ folder and clone this repository (with submodules as described above)
3. Install the conda environment:

conda env create -f environment.yml

4. For the NMPC controller, you must also install Acados — see the NMPC package README for detailed Acados setup instructions
5. In the root of the workspace, build everything:

colcon build --symlink-install

Simulation Setup

1. Run the SITL simulation:

make px4_sitl gazebo-classic

2. Run the Micro XRCE Agent as the bridge between ROS2 and PX4:

MicroXRCEAgent udp4 -p 8888

Running the Quadrotor NR Flow Controller

1. In another terminal, source the workspace and activate your conda environment:

source install/setup.bash
conda activate <your_env_name>

2. Run the controller:

# Simulation — circle trajectory
ros2 run newton_raphson_px4 run_node --platform sim --trajectory circle_horz

# Hardware — helix trajectory with logging
ros2 run newton_raphson_px4 run_node --platform hw --trajectory helix --log

# Hover mode 3, double speed, with yaw spin
ros2 run newton_raphson_px4 run_node --platform sim --trajectory hover --hover-mode 3 --double-speed --spin

NR Flow CLI Options

Flag	Description
--platform {sim,hw}	Target platform (required)
--trajectory {hover,yaw_only,circle_horz,...}	Trajectory type (required)
--hover-mode {1..8}	Hover sub-mode (1-4 for hardware)
--log	Enable CSV data logging
--log-file NAME	Custom log filename
--double-speed	2x trajectory speed
--short	Short variant (fig8_vert)
--spin	Enable yaw rotation
--flight-period SEC	Custom flight duration
--ff	Feedforward marker (only valid with fig8_contraction)

Running the Quadrotor NMPC Controller

1. In another terminal, source the workspace and activate your conda environment:

source install/setup.bash
conda activate <your_env_name>

2. Run the controller:

# Simulation — figure-8 trajectory
ros2 run nmpc_acados_px4 run_node --platform sim --trajectory fig8_horz

# Hardware — helix trajectory with logging
ros2 run nmpc_acados_px4 run_node --platform hw --trajectory helix --log

The NMPC CLI options are the same as the NR Flow options listed above.

---

Blimp Code Instructions

The blimp package (blimp_mpc_fbl_nr) includes MPC, FBL (feedback linearization via CBF), and NR Flow controllers for both simulation and hardware.

Running the Blimp Simulator

1. In a terminal, source the workspace and activate your conda environment:

source install/setup.bash
conda activate <your_env_name>

2. Run a simulation using the run_blimp_sim entry point with a controller/trajectory key and a log filename:

ros2 run blimp_mpc_fbl_nr run_blimp_sim <controller_key> <logfile>

Available Controller Keys for Simulation/Hardware

NR Flow (nr): hardware_nr_circle_horz, hardware_nr_circle_vert, hardware_nr_fig8_horz, hardware_nr_fig8_vert_short, hardware_nr_fig8_vert_tall, hardware_nr_circle_horz_spin, hardware_nr_helix, hardware_nr_helix_spin

MPC (mpc): hardware_mpc_circle_horz, hardware_mpc_circle_vert, hardware_mpc_fig8_horz, hardware_mpc_fig8_vert_short, hardware_mpc_fig8_vert_tall, hardware_mpc_circle_horz_spin, hardware_mpc_helix, hardware_mpc_helix_spin

FBL (fbl): hardware_fbl_circle_horz, hardware_fbl_circle_vert, hardware_fbl_fig8_horz, hardware_fbl_fig8_vert_short, hardware_fbl_fig8_vert_tall, hardware_fbl_circle_horz_spin, hardware_fbl_helix, hardware_fbl_helix_spin

Example

# Run NR Flow on a horizontal circle trajectory
ros2 run blimp_mpc_fbl_nr run_blimp_sim hardware_nr_circle_horz log1.log

# Run MPC on a helix trajectory
ros2 run blimp_mpc_fbl_nr run_blimp_sim hardware_mpc_helix log1.log

# Run FBL on a figure-8 trajectory
ros2 run blimp_mpc_fbl_nr run_blimp_sim hardware_fbl_fig8_horz log1.log

Individual Blimp Entry Points

Each controller/trajectory combination also has a dedicated entry point:

NR Flow:

ros2 run blimp_mpc_fbl_nr run_wardi_circle_horz
ros2 run blimp_mpc_fbl_nr run_wardi_circle_vert
ros2 run blimp_mpc_fbl_nr run_wardi_fig8_horz
ros2 run blimp_mpc_fbl_nr run_wardi_fig8_vert_short
ros2 run blimp_mpc_fbl_nr run_wardi_fig8_vert_tall
ros2 run blimp_mpc_fbl_nr run_wardi_circle_horz_spin
ros2 run blimp_mpc_fbl_nr run_wardi_helix
ros2 run blimp_mpc_fbl_nr run_wardi_helix_spin

NLMPC:

ros2 run blimp_mpc_fbl_nr run_nlmpc_circle_horz
ros2 run blimp_mpc_fbl_nr run_nlmpc_circle_vert
ros2 run blimp_mpc_fbl_nr run_nlmpc_fig8_horz
ros2 run blimp_mpc_fbl_nr run_nlmpc_fig8_vert_short
ros2 run blimp_mpc_fbl_nr run_nlmpc_fig8_vert_tall
ros2 run blimp_mpc_fbl_nr run_nlmpc_circle_horz_spin
ros2 run blimp_mpc_fbl_nr run_nlmpc_helix
ros2 run blimp_mpc_fbl_nr run_nlmpc_helix_spin

CBF (FBL):

ros2 run blimp_mpc_fbl_nr run_cbf_circle_horz
ros2 run blimp_mpc_fbl_nr run_cbf_circle_vert
ros2 run blimp_mpc_fbl_nr run_cbf_fig8_horz
ros2 run blimp_mpc_fbl_nr run_cbf_fig8_vert_short
ros2 run blimp_mpc_fbl_nr run_cbf_fig8_vert_tall
ros2 run blimp_mpc_fbl_nr run_cbf_circle_horz_spin
ros2 run blimp_mpc_fbl_nr run_cbf_helix
ros2 run blimp_mpc_fbl_nr run_cbf_helix_spin

---

Citing this Work

Please see the arxiv preprint here

Authors

Evanns G. Morales-Cuadrado, Luke Baird, Yorai Wardi, Samuel Coogan