CubeSat AOCS Mini Project

This repository contains a small side project built to explore some basic Attitude and Orbit Control System concepts for a CubeSat-type spacecraft.

The project is intentionally compact. Instead of trying to build a full software framework, I kept it as two standalone Python scripts:

• one for attitude dynamics and control
• one for attitude estimation with an Extended Kalman Filter

The goal is not to present flight-ready software, but to show practical interest in AOCS and a working understanding of some of the main ideas: rigid-body dynamics, quaternion-based control, noisy measurements, and state estimation.

Project contents

.
├── adcs_simulation.py
├── ekf_estimation.py
├── README.md
├── requirements.txt
├── plots/
└── docs/

What each script does

adcs_simulation.py

This script simulates a simple 3-axis CubeSat attitude control problem.

It includes:

• rigid-body rotational dynamics
• quaternion attitude propagation
• quaternion PD control
• small disturbance torques
• plots of:
  • attitude error
  • angular velocity
  • control torque
  • quaternion components

The purpose of this part is to show a simple closed-loop attitude regulation problem.

ekf_estimation.py

This script simulates a simple 1-axis attitude estimation problem.

It includes:

• a 1-axis rotational model
• simulated gyro and sun-sensor-like measurements
• an EKF for estimating:
  • angle
  • angular velocity
  • gyro bias
• plots of:
  • true vs estimated angle
  • true vs estimated angular velocity
  • true vs estimated gyro bias
  • angle estimation error with covariance bounds

The purpose of this part is to show a simple sensor-fusion and estimation problem without making the code too heavy.

Why the project is split this way

The control simulation is done in 3 axes because attitude control is much more meaningful there, especially with quaternions. The estimator is kept to 1 axis on purpose. A full 3-axis estimator would be more realistic, but it would also make the project much larger and less transparent. For this repo, I preferred a smaller implementation that is easy to read and explain.

How to run

Run the control simulation:

python adcs_simulation.py

Run the estimation simulation:

python ekf_estimation.py

Both scripts save figures in the plots/ directory.

What this project is meant to show

This repository is mainly a learning and portfolio project. It is meant to show that I took the initiative to experiment with topics that are relevant to entry-level AOCS and GNC work, including:

• rotational dynamics
• quaternion kinematics
• basic feedback control
• disturbance rejection
• sensor modelling
• Extended Kalman Filtering
• gyro bias estimation

Main simplifications

This is a deliberately simplified project. In particular:

• orbit propagation is not modelled
• the control case uses direct torque rather than a detailed actuator model
• the disturbance torques are simple analytical functions
• the EKF example is 1-axis rather than full 3-axis
• environmental models are very basic

These simplifications are intentional. The aim was to build something small, readable, and functional rather than a large unfinished simulator.

Possible next steps

Some natural extensions would be:

• reaction wheel or magnetorquer modelling
• a 3-axis attitude estimator
• more realistic disturbance models
• actuator saturation and sensor update-rate effects
• Monte Carlo runs for sensitivity analysis
• comparison between different control laws