1: \begin{abstract}

2: Simultaneously accurate and reliable tracking control for quadrotors in complex dynamic environments is challenging. {The chaotic nature of} aerodynamics, derived from drag forces and moment variations, {makes precise identification} difficult. {Consequently}, {many existing} quadrotor tracking systems treat {these aerodynamic effects} as simple `disturbances' in conventional control approaches. We propose a novel {and} interpretable trajectory tracker integrating a {d}istributional Reinforcement Learning {(RL)} disturbance estimator for unknown aerodynamic effects with a Stochastic Model Predictive Controller (SMPC). {Specifically, t}he proposed estimator `Constrained Distributional R{E}inforced{-D}isturbance{-}estimator' (ConsDRED) {effectively} identifies uncertainties between {the} true and estimated values of aerodynamic effects. Control parameterization employs {s}implified {a}ffine {d}isturbance {f}eedback to {ensure} convexity, which is seamlessly integrated with {the} SMPC. We theoretically guarantee that ConsDRED achieves an optimal global convergence rate, and sublinear rates if constraints are violated with certain error decreases as neural network dimensions increase. To demonstrate practicality, we show convergent training{,} in simulation and real-world experiments, and empirically verify that ConsDRED is less sensitive to hyperparameter settings compared with canonical constrained RL. {O}ur system {substantially} improves accumulative tracking errors by at least 70\%{,} compared with the recent art. Importantly, the proposed ConsDRED-SMPC {framework} balances the trade-off between pursuing high performance and obeying conservative constraints for practical implementations\footnote{Our code and all research artefacts will be made openly available upon acceptance of our manuscript for publication. Video figures in support are also available: \url{https://github.com/Alex-yanranwang/ConsDRED-SMPC}.}.

3: \end{abstract}

4: