دانشمندای دانشگاه TU Delft هلند یه حرکت خفن زدن و روشی ساختن که چندتا پهپاد با هم یه بارِ آویزون رو بدون نیاز به هوش مصنوعی، یادگیری ماشین یا حتی سنسور روی خودِ بار جابه‌جا کنن

Most multi-drone systems struggle with one thing: coordination under real-world constraints. A paper in Science Robotics from TU Delft proposes a model-based approach that lets multiple quadrotors jointly move and orient a cable-suspended load. BUT: without relying on learning or external sensors on the payload. Key results reported: > More than 8× acceleration of the load compared to previous state-of-the-art methods > Full obstacle avoidance while maintaining internal collision safety > Robust performance even when the load’s inertia model is uncertain > No sensors needed on the load itself Instead of data-driven models, they use a centralized trajectory optimizer that resamples past trajectories for smooth updates and stability. Each drone then executes its reference locally using onboard INDI control. It’s a reminder that not all progress in robotics comes from scaling models… some of it still comes from getting the physics and coordination right. 📍Paper: https://www.science.org/doi/10.1126/scirobotics.adu8015 Video: https://www.youtube.com/watch?v=FBWN-rTK1YU —— Weekly robotics and AI insights. Subscribe free: http://22astronauts.com