{"guid":"66a5b0ea-1518-56f0-9934-1859b1bf29ed","title":"How to prevent the robot uprising - an introduction to MPC based safe reinforcement learning","subtitle":null,"slug":"2025-451-how-to-prevent-the-robot-uprising-an-introduction-to-mpc-based-safe-reinforcement-learning","link":"https://talks.mrmcd.net/2025/talk/URX8PM/","description":"Machine learning in safety critical systems can be... ~a very dumb idea~ a challenge. In this talk I want to introduce a recent framework for a robust safety system for reinforcement learning in safety critical systems based on Model Predictive Control. \nIn short: \u003cb\u003eHow to train a system without blowing up the building in the process\u003c/b\u003e\n\nModel Predictive Control is a recent compute heavy control strategy for autonomous systems. I want to explore its applications to reinforcement learning. A mathematical model of both the enviroment and the system-in-training acts as a supervisor during the training and provides mathematical guarantees for certain bahviour boundaries. To provide some insights I will present my own experiments from my bachelor thesis as well as the research done at University of Lübeck on the topic. If I have time I will try to demonstrate a simple version of the architecture.\n\nhttps://creativecommons.org/licenses/by-sa/4.0/","original_language":"eng","persons":["neolium"],"tags":["451","2025","mrmcd25","C120 - Art. 15 DSGVO","mrmcd25-eng","Darmstadt","mrmcd25","Day 2"],"view_count":93,"promoted":false,"date":"2025-09-13T19:00:00.000+02:00","release_date":"2025-09-13T00:00:00.000+02:00","updated_at":"2026-03-17T22:00:07.581+01:00","length":2240,"duration":2240,"thumb_url":"https://static.media.ccc.de/media/conferences/mrmcd/mrmcd25/451-66a5b0ea-1518-56f0-9934-1859b1bf29ed.jpg","poster_url":"https://static.media.ccc.de/media/conferences/mrmcd/mrmcd25/451-66a5b0ea-1518-56f0-9934-1859b1bf29ed_preview.jpg","timeline_url":"https://static.media.ccc.de/media/conferences/mrmcd/mrmcd25/451-66a5b0ea-1518-56f0-9934-1859b1bf29ed.timeline.jpg","thumbnails_url":"https://static.media.ccc.de/media/conferences/mrmcd/mrmcd25/451-66a5b0ea-1518-56f0-9934-1859b1bf29ed.thumbnails.vtt","frontend_link":"https://media.ccc.de/v/2025-451-how-to-prevent-the-robot-uprising-an-introduction-to-mpc-based-safe-reinforcement-learning","url":"https://api.media.ccc.de/public/events/66a5b0ea-1518-56f0-9934-1859b1bf29ed","conference_title":"MRMCD 2025 - Volle Transparenz","conference_url":"https://api.media.ccc.de/public/conferences/mrmcd25","related":[],"recordings":[{"size":217,"length":2240,"mime_type":"video/webm;codecs=av01","language":"eng","filename":"mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_av1-hd.webm","state":"new","folder":"av1-hd","high_quality":true,"width":1920,"height":1080,"updated_at":"2025-09-13T22:51:20.873+02:00","recording_url":"https://cdn.media.ccc.de/events/mrmcd/mrmcd25/av1-hd/mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_av1-hd.webm","url":"https://api.media.ccc.de/public/recordings/90986","event_url":"https://api.media.ccc.de/public/events/66a5b0ea-1518-56f0-9934-1859b1bf29ed","conference_url":"https://api.media.ccc.de/public/conferences/mrmcd25"},{"size":34,"length":2240,"mime_type":"audio/mpeg","language":"eng","filename":"mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_mp3.mp3","state":"new","folder":"mp3","high_quality":false,"width":0,"height":0,"updated_at":"2025-09-13T22:20:39.941+02:00","recording_url":"https://cdn.media.ccc.de/events/mrmcd/mrmcd25/mp3/mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_mp3.mp3","url":"https://api.media.ccc.de/public/recordings/90977","event_url":"https://api.media.ccc.de/public/events/66a5b0ea-1518-56f0-9934-1859b1bf29ed","conference_url":"https://api.media.ccc.de/public/conferences/mrmcd25"},{"size":20,"length":2240,"mime_type":"audio/opus","language":"eng","filename":"mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_opus.opus","state":"new","folder":"opus","high_quality":false,"width":0,"height":0,"updated_at":"2025-09-13T22:19:35.130+02:00","recording_url":"https://cdn.media.ccc.de/events/mrmcd/mrmcd25/opus/mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_opus.opus","url":"https://api.media.ccc.de/public/recordings/90976","event_url":"https://api.media.ccc.de/public/events/66a5b0ea-1518-56f0-9934-1859b1bf29ed","conference_url":"https://api.media.ccc.de/public/conferences/mrmcd25"},{"size":263,"length":2240,"mime_type":"video/webm","language":"eng","filename":"mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_webm-hd.webm","state":"new","folder":"webm-hd","high_quality":true,"width":1920,"height":1080,"updated_at":"2025-09-13T23:26:26.775+02:00","recording_url":"https://cdn.media.ccc.de/events/mrmcd/mrmcd25/webm-hd/mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_webm-hd.webm","url":"https://api.media.ccc.de/public/recordings/90996","event_url":"https://api.media.ccc.de/public/events/66a5b0ea-1518-56f0-9934-1859b1bf29ed","conference_url":"https://api.media.ccc.de/public/conferences/mrmcd25"},{"size":81,"length":2240,"mime_type":"video/webm","language":"eng","filename":"mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_webm-sd.webm","state":"new","folder":"webm-sd","high_quality":false,"width":720,"height":576,"updated_at":"2025-09-13T23:09:03.762+02:00","recording_url":"https://cdn.media.ccc.de/events/mrmcd/mrmcd25/webm-sd/mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_webm-sd.webm","url":"https://api.media.ccc.de/public/recordings/90990","event_url":"https://api.media.ccc.de/public/events/66a5b0ea-1518-56f0-9934-1859b1bf29ed","conference_url":"https://api.media.ccc.de/public/conferences/mrmcd25"},{"size":88,"length":2240,"mime_type":"video/mp4","language":"eng","filename":"mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_sd.mp4","state":"new","folder":"h264-sd","high_quality":false,"width":720,"height":576,"updated_at":"2025-09-13T22:30:03.815+02:00","recording_url":"https://cdn.media.ccc.de/events/mrmcd/mrmcd25/h264-sd/mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_sd.mp4","url":"https://api.media.ccc.de/public/recordings/90981","event_url":"https://api.media.ccc.de/public/events/66a5b0ea-1518-56f0-9934-1859b1bf29ed","conference_url":"https://api.media.ccc.de/public/conferences/mrmcd25"},{"size":333,"length":2240,"mime_type":"video/mp4","language":"eng","filename":"mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_hd.mp4","state":"new","folder":"h264-hd","high_quality":true,"width":1920,"height":1080,"updated_at":"2025-09-13T21:51:28.291+02:00","recording_url":"https://cdn.media.ccc.de/events/mrmcd/mrmcd25/h264-hd/mrmcd25-451-eng-How_to_prevent_the_robot_uprising_-_an_introduction_to_MPC_based_safe_reinforcement_learning_hd.mp4","url":"https://api.media.ccc.de/public/recordings/90965","event_url":"https://api.media.ccc.de/public/events/66a5b0ea-1518-56f0-9934-1859b1bf29ed","conference_url":"https://api.media.ccc.de/public/conferences/mrmcd25"}]}