neurofly-dataset-generation

Utilities for generating UAV trajectory datasets in RotorPy, adding configurable sensor noise, and plotting the resulting trajectories. The current workflow focuses on a smooth conical spiral flight path and produces CSV files that are useful for filtering, state-estimation, and sensor-fusion experiments.

Purpose

This repository provides three main capabilities:

1. Run a RotorPy simulation of a spiral flight path and export the resulting CSV logs.
2. Augment a simulation CSV with noisy IMU, mocap or GPS-style measurements, plus a drifting dead-reckoned state.
3. Visualize tracking quality and noisy-vs-clean position data.

The code is organized as a small Python package in neurofly_dataset_generation, while the original top-level scripts remain as compatibility entry points:

• main.py: run the simulation and export datasets
• inject_trajectory_noise.py: create a noisy CSV
• visualise_traj.py: plot desired vs followed trajectory
• plot_noisy_position.py: plot clean vs noisy position

The preferred interface is now the package CLI:

python -m neurofly_dataset_generation <command> [options]

Available commands:

• generate
• add-noise
• plot-tracking
• plot-noisy

Dependencies

• Python 3.10+
• numpy
• scipy
• matplotlib
• rotorpy

Dependencies are declared in pyproject.toml and requirements.txt.

Setup

If you already have a working environment with RotorPy installed, you can use it directly. In this workspace there is also a local virtual environment under rotorpy/.

Example setup:

python3 -m venv .venv
. .venv/bin/activate
pip install -r requirements.txt
pip install -e .

If you prefer the existing local environment in this folder:

rotorpy/bin/pip install -e .

Usage

1. Generate a simulation dataset

rotorpy/bin/python main.py --plot-tracking --tracking-output-dir trajectory_plots

Equivalent package CLI:

rotorpy/bin/python -m neurofly_dataset_generation generate --plot-tracking --tracking-output-dir trajectory_plots

This writes:

• basic_usage.csv: full RotorPy state and reference log
• spiral_flight_data.csv: compact IMU-oriented export
• trajectory_plots/: optional tracking plots

Useful flags:

• --no-plot
• --no-animate
• --quiet
• --sensor-csv <path>
• --simulation-csv <path>

2. Inject sensor noise

python3 inject_trajectory_noise.py basic_usage.csv --output basic_usage_noisy.csv --abs-sensor mocap --seed 7

Equivalent package CLI:

python3 -m neurofly_dataset_generation add-noise basic_usage.csv --output basic_usage_noisy.csv --abs-sensor mocap --seed 7

Options:

• --abs-sensor mocap
• --abs-sensor gps
• --abs-sensor none
• --noise-scale <float>
• --mode deadreckon

The generated noisy CSV appends:

• noisy accelerometer and gyro channels
• optional noisy mocap or GPS-like position channels
• noisy wind and rotor-speed channels
• a dead-reckoned drifting state estimate

3. Plot desired vs tracked trajectory

python3 visualise_traj.py basic_usage.csv --output-dir trajectory_plots --show

Equivalent package CLI:

python3 -m neurofly_dataset_generation plot-tracking basic_usage.csv --output-dir trajectory_plots --show

4. Plot clean vs noisy position

python3 plot_noisy_position.py basic_usage_noisy.csv --output-dir trajectory_plots --show

Equivalent package CLI:

python3 -m neurofly_dataset_generation plot-noisy basic_usage_noisy.csv --output-dir trajectory_plots --show

Project Structure

neurofly_dataset_generation/
  cli.py           Command-line entry points
  config.py        Shared defaults
  io_utils.py      CSV and filesystem helpers
  noise.py         Noise injection and dead reckoning
  plotting.py      Visualization helpers
  simulation.py    RotorPy simulation orchestration
  trajectories.py  Spiral trajectory definitions

Notes

• Generated CSV files and plot outputs are ignored by Git by default.
• The local rotorpy/ directory is treated as an environment directory and is also ignored.
• The simulation entry point requires a Python environment where the rotorpy package is installed.