control systems

Conditional Invertible Neural Networks for Data-Driven UAV Control: A 2-D Proof of Concept

arXiv:2607.13703

summary

The paper evaluates conditional invertible neural networks as probabilistic inverse‑dynamics models for controlling a planar X8 multicopter, learning from an INDI teacher and comparing open‑ and closed‑loop performance.

Abstract

We investigate conditional invertible neural networks (cINNs) as probabilistic inverse-dynamics models for multirotor control. For a planar X8 coaxial multicopter, we learn from an incremental nonlinear dynamic inversion (INDI) teacher using rational-quadratic spline coupling and invertible linear mixing. Open-loop reproduction reaches , mean CRPS 0.0915, and log-probability-error correlation . Over 15 closed-loop scenarios, position RMSE matches INDI (9.7 vs. 9.5 m), with 47 percent tracking acceptably; failures separate into attitude divergence under aggressive steps and phase lag under high-frequency references, isolating command bandwidth and data coverage as dominant failure mechanisms.

Topics & keywords

#invertible neural networks#probabilistic modeling#UAV control#inverse dynamics#multirotorconditional invertible neural networkcINNrational-quadratic spline couplingincremental nonlinear dynamic inversionCRPSlog-probability error
Conditional Invertible Neural Networks for Data-Driven UAV Control: A 2-D Proof of Concept · wovepaper