Scaling PX4 for Heavy-Lift Drones: Challenges in Planning and Control - Aniruddha Mallick, TUM

Scaling PX4 for Heavy-Lift Drones: Challenges in Planning and Control - Aniruddha Mallick, TUM This session presents the adaptation and deployment of a PX4-based autonomy stack for large-scale UAVs, focusing on reliable autonomous navigation under real-world constraints. We explore a receding horizon control architecture that integrates global path planning, local trajectory optimization, and low-level trajectory tracking, while accounting for vehicle dynamics, computational delays, and resource limitations. While widely used for small-scale drones, extending these architectures to heavy-lift UAVs introduces significant challenges in handling dynamics, control limits, and motion agility. The session draws on work from simulation, hardware-in-the-loop testing, and early system integration on scaled down prototypes. This talk explores the key adaptations needed to enable real-time planning and control in dynamic environments using PX4 and ROS2. The goal is to share insights and inspire discussion on extending the open-source PX4 ecosystem for high-payload autonomous flight.