Congratulations, Alp Can Karacakol and team! 🎉🎉🎉
📢 Excited to share our new article in Nature Communications: “Data-driven design of shape-programmable magnetic soft materials” We developed a data-driven computational framework to automatically design magnetic soft materials, optimizing their shape, magnetic programming, and material composition. Using machine-learning-guided simulations, our method rapidly explores an enormous design space, uncovering complex 2D and 3D shape-morphing behaviors far beyond human intuition. Experimental validation confirms reliable simulation-to-reality transfer, paving the way for advanced soft robotics and adaptive wearable technologies. 🌐 Open Science: To drive progress in the community, we've openly shared: 1️⃣ All developed algorithm and simulation tools 2️⃣ Comprehensive dataset of 15 million designs to support future machine learning research in soft robotics 3️⃣ 97 pages of technical details, ensuring reproducibility and ease of benchmarking 🔎 Dive deeper: o Read the article: https://lnkd.in/eMd3EVCM o Explore the code: https://lnkd.in/eHgRQnjx o Access the dataset: https://lnkd.in/eX_wTg5m 👏 Huge thanks and kudos to the amazing team: Yunus Alapan, Sinan Ozgun Demir, Metin Sitti This work was funded by the Max Planck Society and European Research Council (ERC) Advanced Grant SoMMoR project with grant no: 834531. I also personally thank to the Max Planck ETH Center for Learning Systems for the funding during a part of this work. #robotics #magneticrobotics #softrobotics #machinelearning #datadrivendesign 🧩 Workflow Overview (from attached figure): (A-C) Defining desired behaviors and conditions; (D-E) exploring the design space via simulations guided by machine learning; (F) identifying optimal designs; and (G) validating the chosen design experimentally.