Autonomous Drone Racing

Agile flight tasks such as drone racing and pursuit-evasion require strong acceleration and precise turns, but the available thrust changes as the battery discharges and voltage drops under load. Conservative command limits make this variation easier to tolerate, at the cost of unused performance. We investigate how learned controllers can use that additional thrust while retaining the flight controller's voltage compensation and rate control. Our training simulator couples an identified load-transient battery model to rotor dynamics and firmware saturation. The feedforward policy receives filtered voltage during both training and deployment. Controlled ablations distinguish the benefit of a larger thrust-command range from that of voltage information. On a 38 g Crazyflie Brushless, the resulting policy reduces circle tracking error by 49% relative to stock-authority RL at 3.84 m/s, while preserving easy-task precision. Mean 20-lap race time decreases from 106.22 s to 95.24 s. Compared with a voltage-blind policy with the same increased authority, hardware error and race time are lower by 15.3% and 4.5%, respectively. In simulation, replacing the policy's voltage input with a recording from a different battery condition worsens hard-circle tracking, with a smaller, voltage-dependent effect in racing. Together, these results show where a simple voltage input complements existing actuator compensation in aggressive learned flight.

Quadrotor racing demands aggressive attitude and progress control while passing through every gate, and conventional quadrotor MPCC formulations state the prediction model in inertial coordinates and the attitude error in the body frame. We present a Dual-Quaternion Model Predictive Contouring Control (DQ-MPCC) for quadrotor racing in which the pose is a unit dual quaternion and the contouring errors are projected onto the tangent space of the dual quaternion manifold, expressed in the desired body frame: the same rigid-body dynamics as the conventional model, in unified pose-twist coordinates in the body frame. We compare DQ-MPCC against a baseline MPCC through Monte Carlo software-in-the-loop simulations and real-world racing on an eight-gate circuit of 11x4.5x3.65 m. With the same gains in simulation and hardware, DQ-MPCC keeps every crossing of its completed flights within the prescribed geometric margin, whereas the baseline exceeds it, its median worst-gate offset growing by 71.5% sim-to-real against a 10.1% decrease for DQ-MPCC. Among the configurations that keep every simulated gate crossing within the geometric margin, DQ-MPCC also reduces the minimum lap time by 6.7%, and by 10.5% in the real-world flights, while running onboard at 100 Hz.

Autonomous drone racing is a fundamentally challenging regime for autonomous aerial robots, requiring time-optimal control while operating under persistent actuation saturation. While reinforcement learning (RL) has achieved human-level performance in this domain, current methods fail to generalize; policies trained on specific environments often crash immediately in unseen configurations. This failure reflects the intrinsic difficulty of zero-shot generalization in agile flight, arising from high-dimensional task variation and the tight coupling between safety and performance at high speeds. Existing approaches that improve generalization impose a substantial cost on flight speed: control policies must significantly degrade performance to achieve even modest levels of generalization. In this work, we propose a framework for zero-shot generalization in agile flight for RL-based drone racing. By combining task-aware switching based on learning progress with a physically informed procedural track generator, the framework produces a fast and robust generalist policy without test-time adaptation. Our method achieves strong zero-shot performance across a wide range of unseen racetracks in the real world, demonstrating a 7.4x improvement in generalization over the state-of-the-art approaches, while maintaining competitive racing speeds. We validate our method's results in both simulation and real-world settings, including a challenging vision-based, end-to-end control setting that operates without explicit state estimation, where all prior approaches fail to generalize.

In the literature, actor-critic model predictive control (AC-MPC) integrates MPC with reinforcement learning to enable high-performance control of complex dynamical systems. However, its differentiable MPC layer requires repeatedly solving an optimization problem in both the forward and backward passes, leading to substantial training and inference latency. This paper tackles this bottleneck introducing a CUDA-accelerated variant that significantly reduces end-to-end execution time while preserving the control performance of the baseline formulation. Simulation results on an agile drone racing task show that our approach achieves state-of-the-art lap times and near-limit dynamic behaviour with markedly reduced training and inference time.

Autonomous systems have achieved superhuman performance in isolation or simulation, yet they remain brittle in shared, dynamic real-world spaces. This failure stems from the dominant single-agent paradigm for physical applications, where other actors are ignored or treated as environmental noise, preventing effective coordination. Here we show that multi-agent reinforcement learning provides the essential safety scaffolding required for real-world interaction. Using high-speed quadrotor racing as a high-stakes testbed, we train agents to navigate complex aerodynamic interactions and strategic maneuvering with a variable number of racers. Through league-based self-play, agents evolve sophisticated anticipatory behaviors, including proactive collision avoidance, overtaking, and handling multi-agent physical interactions, including aerodynamic downwash. Our agents outperform a champion-level human pilot in multi-player races at speeds exceeding 22 m/s, while simultaneously reducing collision rates by 50 % compared to state-of-the-art single-agent baselines. Crucially, training with diverse artificial agents enables zero-shot generalization to safer human interaction. These results suggest that the path to robust robotic co-existence lies not in isolated safety constraints, but in the rigorous demands of multi-agent interaction. Multimedia materials are available at: https://rpg.ifi.uzh.ch/marl

Robotic tasks are typically specified by a tuple of factors, such as the object to be grasped, the obstacles to be avoided, the color of the target, and so on. Collecting expert demonstrations for every combination of factor values grows combinatorially. We present factored diffusion policies: a single shared diffusion network trained with per-factor null-token dropout, whose score decomposes additively across factors at inference. Under approximate conditional independence between factors given the action-observation pair, this composition approximates the true joint score with a bounded uniform error, reducing the training-task budget from a product of factor cardinalities to a sum. A trajectory-tube certificate chains this score-level bound through the reverse-time sampling ODE and a contracting tracking controller into a closed-loop state-trajectory tube whose radius factors into an ODE-sensitivity constant and a per-factor score-error budget. Unlike compositional-diffusion methods for control that combine separately trained networks, we use one shared network. Drone racing experiments confirm both the generalization bound and the certificate. On state-based multi-gate racing, the factored policy passes 90% of held-out gates -- matching an oracle -- while a K-network composition baseline collapses to 3%; on vision-based single-gate traversal, it transfers zero-shot to an unseen venue with +11.7pp success-rate gain and 2.4X crash-rate reduction.

Reinforcement learning (RL) often struggles in real-world tasks with high-dimensional state spaces and long horizons, where sparse or fixed rewards severely slow down exploration and cause agents to get trapped in local optima. This paper presents a fuzzy logic based reward shaping method that integrates human intuition into RL reward design. By encoding expert knowledge into adaptive and interpreable terms, fuzzy rules promote stable learning and reduce sensitivity to hyperparameters. The proposed method leverages these properties to adapt reward contributions based on the agent state, enabling smoother transitions between fast motion and precise control in challenging navigation tasks. Extensive simulation results on autonomous drone racing benchmarks show stable learning behavior and consistent task performance across scenarios of increasing difficulty. The proposed method achieves faster convergence and reduced performance variability across training seeds in more challenging environments, with success rates improving by up to approximately 5 percent compared to non fuzzy reward formulations.

Autonomous drone racing demands robust real-time localization under extreme conditions: high-speed flight, aggressive maneuvers, and payload-constrained platforms that often rely on a single camera for perception. Existing visual SLAM systems, while effective in general scenarios, struggle with motion blur and feature instability inherent to racing dynamics, and do not exploit the structured nature of racing environments. In this work, we present a dual pose-graph architecture that fuses odometry with semantic detections for robust localization. A temporary graph accumulates multiple gate observations between keyframes and optimizes them into a single refined constraint per landmark, which is then promoted to a persistent main graph. This design preserves the information richness of frequent detections while preventing graph growth from degrading real-time performance. The system is designed to be sensor-agnostic, although in this work we validate it using monocular visual-inertial odometry and visual gate detections. Experimental evaluation on the TII-RATM dataset shows a 56% to 74% reduction in ATE compared to standalone VIO, while an ablation study confirms that the dual-graph architecture achieves 10% to 12% higher accuracy than a single-graph baseline at identical computational cost. Deployment in the A2RL competition demonstrated that the system performs real-time onboard localization during flight, reducing the drift of the odometry baseline by up to 4.2 m per lap.

Autonomous drone racing in complex environments requires agile, high-speed flight while maintaining reliable obstacle avoidance. Differentiable-physics-based policy learning has recently demonstrated high sample efficiency and remarkable performance across various tasks, including agile drone flight and quadruped locomotion. However, applying such methods to drone racing remains difficult, as key objective like gate traversal are inherently hard to express as smooth, differentiable losses. To address these challenges, we propose DiffRacing, a novel vector field-augmented differentiable policy learning framework. DiffRacing integrates differentiable losses and vector fields into the training process to provide continuous and stable gradient signals, balancing obstacle avoidance and high-speed gate traversal. In addition, a differentiable Delta Action Model compensates for dynamics mismatch, enabling efficient sim-to-real transfer without explicit system identification. Extensive simulation and real-world experiments demonstrate that DiffRacing achieves superior sample efficiency, faster convergence, and robust flight performance, thereby demonstrating that vector fields can augment traditional gradient-based policy learning with a task-specific geometric prior.

Autonomous drone racing requires the tight coupling of perception, planning, and control under extreme agility. However, recent approaches typically rely on precomputed spatial reference trajectories or explicit 6-DoF gate pose estimation, rendering them brittle to spatial perturbations, unmodeled track changes, and sensor noise. Conversely, end-to-end learning policies frequently overfit to specific track layouts and struggle with zero-shot generalization. To address these fundamental limitations, we propose a fully onboard, vision guided optimal control framework that enables reference-free agile flight through arbitrarily placed and oriented gates. Central to our approach is Gate-SDF, a novel, implicitly learned neural signed distance field. Gate-SDF directly processes raw, noisy depth images to predict a continuous spatial field that provides both collision repulsion and active geometric guidance toward the valid traversal area. We seamlessly integrate this representation into a sampling-based Model Predictive Path Integral (MPPI) controller. By fully exploiting GPU parallelism, the framework evaluates these continuous spatial constraints across thousands of simulated trajectory rollouts simultaneously in real time. Furthermore, our formulation inherently maintains spatial consistency, ensuring robust navigation even under severe visual occlusion during aggressive maneuvers. Extensive simulations and real-world experiments demonstrate that the proposed system achieves high-speed agile flight and successfully navigates unseen tracks subject to severe unmodeled gate displacements and orientation perturbations. Videos are available at https://zhaofangguo.github.io/vision_guided_mppi/

Autonomous drone racing (ADR) demands state estimation that is simultaneously computationally efficient and resilient to the perceptual degradation experienced during extreme velocity and maneuvers. Traditional frameworks typically rely on conventional visual-inertial pipelines with loosely-coupled gate-based Perspective-n-Points (PnP) corrections that suffer from a rigid requirement for four visible features and information loss in intermediate steps. Furthermore, the absence of GNSS and Motion Capture systems in uninstrumented, competitive racing environments makes the objective evaluation of such systems remarkably difficult. To address these limitations, we propose ADR-VINS, a robust, monocular visual-inertial state estimation framework based on an Error-State Kalman Filter (ESKF) tailored for autonomous drone racing. Our approach integrates direct pixel reprojection errors from gate corners features as innovation terms within the filter. By bypassing intermediate PnP solvers, ADR-VINS maintains valid state updates with as few as two visible corners and utilizes robust reweighting instead of RANSAC-based schemes to handle outliers, enhancing computational efficiency. Furthermore, we introduce ADR-FGO, an offline Factor-Graph Optimization framework to generate high-fidelity reference trajectories that facilitate post-flight performance evaluation and analysis on uninstrumented, GNSS-denied environments. The proposed system is validated using TII-RATM dataset, where ADR-VINS achieves an average RMS translation error of 0.143 m, while ADR-FGO yields 0.060 m as a smoothing-based reference. Finally, ADR-VINS was successfully deployed in the A2RL Drone Championship Season 2, maintaining stable and robust estimation despite noisy detections during high-agility flight at top speeds of 20.9 m/s. We further utilize ADR-FGO for post-flight evaluation in uninstrumented racing environments.

Autonomous drone racing has attracted increasing interest as a research topic for exploring the limits of agile flight. However, existing studies primarily focus on obstacle-free racetracks, while the perception and dynamic challenges introduced by obstacles remain underexplored, often resulting in low success rates and limited robustness in real-world flight. To this end, we propose a novel vision-based curriculum reinforcement learning framework for training a robust controller capable of addressing unseen obstacles in drone racing. We combine multi-stage cu rriculum learning, domain randomization, and a multi-scene updating strategy to address the conflicting challenges of obstacle avoidance and gate traversal. Our end-to-end control policy is implemented as a single network, allowing high-speed flight of quadrotors in environments with variable obstacles. Both hardware-in-the-loop and real-world experiments demonstrate that our method achieves faster lap times and higher success rates than existing approaches, effectively advancing drone racing in obstacle-rich environments. The video and code are available at: https://github.com/SJTU-ViSYS-team/CRL-Drone-Racing.