[Paper] 8th IFAC NMPC2024 Award]

🏎 In our paper, we propose practical approaches that will hopefully accelerate the adoption of NMPC in industry, particularly for safety-critical systems such as autonomous driving like we are doing it at Siemens Digital Industries Software in collaboration with KU Leuven.

🎯 𝐖𝐡𝐚𝐭 𝐚𝐫𝐞 𝐰𝐞 𝐚𝐢𝐦𝐢𝐧𝐠 𝐟𝐨𝐫? Safety and real-time computation without compromising performance!

⚙ 𝐖𝐡𝐚𝐭 𝐝𝐨𝐞𝐬 𝐭𝐡𝐞 𝐰𝐨𝐫𝐤 𝐨𝐟𝐟𝐞𝐫 𝐢𝐧 𝐭𝐞𝐫𝐦𝐬 𝐨𝐟 𝐬𝐚𝐟𝐞𝐭𝐲? 1️⃣ The introduction of 𝗥𝗘𝗦𝗔𝗙𝗘/𝗖𝗢𝗟, a novel, scalable and accurate approach to transform the continuous-time NMPC problem into a numerical optimization problem. 𝗥𝗘𝗦𝗔𝗙𝗘/𝗖𝗢𝗟 offers guaranteed continuous-time constraint satisfaction while still being at least 5x faster than state-of-practice methods. 2️⃣ The use of control barrier functions (CBF) constraints to allow earlier reaction to upcoming obstacles in urban driving autonomous driving, even with a limited planning horizon. CBFs allow to design optimal decision making such that safety constraints could be satisfied over infinite time ahead.

Bonus: Real-time NMPC application and memory efficiency due to a scalable and tractable RESAFE/COL if you’re aiming for embedded hardware deployment!

✅Does it work in practice? You’ll be the judge as we put it in an automated driving use case designed with Simcenter Amesim and Prescan (can you spot the Munich streets?) https://lnkd.in/erwWT9J8

A big thanks to the co-authors Panagiotis (Panos) Patrinos (KU Leuven), Toshiyuki Ohtsuka (Kyoto University) and Son Tong for their invaluable contributions.

📖Interested in more details? Check the paper preprint below: https://lnkd.in/ekPf-e4T