High-speed near-ground flight presents critical challenges for large-inertia UAVs carrying payloads, including complex obstacles and communication-denied environments. Unlike agile small drones, these platforms require both rapid path planning and strict adherence to trajectory tracking constraints for safe obstacle avoidance. This paper proposes a two-stage autonomous navigation framework tailored for large-inertia UAVs. The framework integrates: (1) an enhanced LiDAR model with physical optical noise for improved simulation fidelity; (2) an ESDF + OctoMap dual-map construction method supporting global search and local optimization; and (3) a global BIT* planner combined with a B-spline local optimizer embedding dynamic, smoothness, and tracking accuracy constraints to ensure path feasibility and trackability. Simulation results demonstrate an average planning time of 0.86 ms, outperforming NAVIGATION, Informed RRT*, MPC Planner, and ESDF Optimization by 29.6–52.0%, with a 100% obstacle avoidance success rate and trajectory tracking RMSE of 0.28 m over a 350 m flight distance, along with strong parameter and noise robustness. Actual flight tests on a 9.4 kg quadrotor UAV confirm the algorithm’s effectiveness in map construction, path planning, and obstacle avoidance in environments with 15 obstacles, while maintaining computational overhead suitable for onboard deployment. These results establish the proposed framework as an effective solution for high-speed autonomous navigation of large-inertia UAVs in complex near-ground environments.
Xiong et al. (Fri,) studied this question.