Closed-Form Continuous-Time Neural Networks for UAV Robot Position Estimation
September 25, 2026
A closed-form continuous-time (CfC) neural network is proposed to estimate end-effector position for aerial continuum manipulators under aerodynamic disturbances. The model outperforms standard MLP and GRU architectures in predicting 3D position residuals during UAV free-hovering conditions.
HOW THIS AFFECTS YOU
●
builderYou can apply CfC architectures to improve state estimation in robotics operating in dynamic, high-disturbance environments.
●
researcherThis provides a comparative baseline for temporal neural networks in aerial manipulation tasks.