Build autonomous flight from equations that survive contact with hardware.
The programme connects attitude mathematics, noisy sensors, state estimation, rigid-body dynamics, control, embedded execution, and physical test rigs.
Each layer is developed as an inspectable contract before it becomes an autonomy claim.
Flight control is a chain of models, not one PID loop
Errors compound across frames, sensors, estimators, dynamics, control laws, timing, and actuators.
Frames + attitude
Vectors, Euler angles, DCMs, quaternions, and interpolation define orientation consistently.
Sensors
IMU, magnetometer, barometer, calibration, noise, bias, and timing define what can be observed.
State estimation
Complementary and Kalman-family filters turn measurements into attitude, velocity, position, and uncertainty.
Vehicle dynamics
Rigid-body motion, rotor forces, torque, actuator lag, gravity, and integration define the plant.
Control
PID, LQR, MPC, robust control, mixing, constraints, and saturation turn state error into commands.
Embedded + validation
Zephyr targets, telemetry, simulation, rigs, and thrust measurement close the loop with hardware.
Autonomy matures one verified contract at a time
The diagram separates demonstrated mathematics and software from in-progress and planned work. It is a maturity map, not a finished-product claim.
The mathematical stack is split into inspectable systems
These repositories are public for inspection. Check each repository for its own reuse terms.
attitudeMathLibrary
Quaternion, Euler, DCM, vector, and SLERP mathematics.
stateEstimation
Complementary filters, EKF, bias, magnetometer, and altitude fusion.
dynamic_models + rotorDynamics
Six-DOF rigid-body motion, linearization, rotor thrust, torque, and induced velocity.
controlSystems
PID, LQR, MPC, H-infinity, mixers, and closed-loop simulation.
inertial_navigation_system
Estimator, controller, dynamics, and hardware abstraction integration.
AeroDynControlRig + ThrustStand
Visualization, tuning, telemetry, and physical propulsion measurement.
What exists now, and what remains work
Roadmap language is not presented as measured capability.
Mathematics and software components
Core math, models, estimators, controllers, embedded experiments, and rigs are publicly inspectable.
Integrated validation
The work is converging toward consistent simulation, hardware-in-the-loop, and controlled physical tests.
Autonomous systems
Higher-level autonomy follows reliable estimation, control, timing, and safety evidence.
Production flight certification
The current work is engineering research, not a certified flight-control product.
Antshiv Robotics builds intelligent systems we can understand all the way down, from mathematical kernels and CPU runtimes to simulation, flight control, and autonomous machines.