München AI, Machine Learning and Computer Vision Meetup

Sep 09, 2026
9:00 AM - 10:00 AM PST
Online. Register for the Zoom!

Physical AI still relies on familiar computer vision tasks—detection, segmentation, depth, tracking. What's changed is the data unit: no longer a single image and label, but an episode—a dozen sensors ticking on independent clocks for minutes, with no frame boundaries.

Most computer vision tooling assumes the old unit and breaks on the new one.

That's why Physical AI teams end up with buckets of .mcap files nobody can characterize. Recording is cheap, so logs pile up faster than anyone curates them. Ask what's actually in there—which tasks, which conditions, how many failures and of what kind—and the honest answer is usually a shrug.
MCAP has been ROS 2's default log format since Iron, and as of FiftyOne 1.19 it opens natively: cameras, LiDAR, GPS, IMU, and logs on one shared timeline, alongside your images and video.

We'll tackle quality, the harder half: what smoothness, sensor-health, and outlier metrics actually measure, where each falls short, and how to turn a score into a defensible decision.

You'll leave knowing how to load your own recordings, query a whole corpus instead of a single file, and which quality signals to trust for which job.

Sep 17, 2026
9:00 AM - 11:00 AM PST
Online. Register for the Zoom!

AI for Autonomous Driving: From Data to Decisions

Building reliable automated driving systems is as much a data and engineering challenge as a modeling one. In this talk, Tin will share perspectives from his work at Porsche AG on applying modern AI methods across the autonomous driving development process, from making sense of large-scale driving data to understanding and evaluating how AI-based systems behave on the road. He'll discuss lessons learned from real-world development, where today's approaches shine, and where hard problems remain for the ADAS and AV community.

About the Speaker

Tin Stribor Sohn is a PhD Student at Porsche AG and Karlsruhe Institute of Technology in the area of Foundation Models for Scenario Understanding and Decision Making in Autonomous Robotics, Tech Lead at Data Driven Engineering for Autonomous Driving, Prior: Master in CS at University of Tuebingen with focus on Computer Vision and Deep Learning and co-founder of a software company for smart EV charging

Advancing ADAS and Autonomous Vehicle Development with Multimodal Data

ADAS and autonomous vehicle systems rely on increasingly complex data from cameras, video, LiDAR, radar, and other sensor streams. In this session, Murilo will introduce Voxel51 and explore how the latest multimodal capabilities in FiftyOne help teams bring these data sources together to better understand their datasets and model behavior. He’ll discuss how unified workflows for visualization, search, curation, and evaluation can help ADAS and AV teams uncover challenging scenarios, investigate model failures, and build safer, more reliable autonomous systems.

About the Speaker

Murilo Gustineli is a Machine Learning Engineer at Voxel51 working at the intersection of representation learning and computer vision. He holds an M.S. in Computer Science from Georgia Tech, where he co-leads the DS@GT Applied Research & Competitions group, advancing machine learning research through competitive challenges and peer-reviewed publications.

From Survey-Grade Maps to Physical AI: Scaling Real-World Data for Training and Simulation

Physical AI systems are increasingly constrained not by model architectures, but by the availability of scalable, high-fidelity real-world data. This talk explores how Dynamic Map Platform transforms survey-grade road assets collected across 1.8 million km of roads worldwide into training- and simulation-ready datasets, including point clouds, imagery, HD maps, road surface models, and 3D Gaussian Splatting representations.

We will discuss why geometric accuracy, semantic understanding, and real-world diversity are critical to building robust autonomous driving systems. Attendees will learn how real-world geospatial data can be structured and scaled for AI training and simulation workflows.

About the Speaker

Ryoto Miyake is a Software Engineer at Dynamic Map Platform, where he works on transforming large-scale geospatial data into AI-ready datasets for training, simulation, and validation, such as HD maps and 3D Gaussian Splatting. With a background in transportation engineering, he works closely with automotive manufacturers and industry partners to bridge large-scale real-world mapping data with next-generation AI and mobility systems.

Sep 22, 2026
9:00 AM - 10:00 AM Pacific
Online. Register for the Zoom!

Ask it to find and remove duplicate images, run object detection and surface low-confidence predictions, or evaluate a model and summarize where it fails, and the agent handles execution end to end.

We will also walk through the newest capabilities shipping with this release, including code generation and plugin generation. You will see how to go from a conversational prompt to a custom dashboard, visualization, or full FiftyOne application, and how to package multi-step workflows as reusable skills the agent can call on demand.

Because the agent runs inside FiftyOne's secure guardrails, teams keep full control. Connect your preferred models from over 100 LLM providers, route requests through your own enterprise gateway, and maintain audit logging and user attribution for every action the agent takes.

Seats are limited, Pre-registration is mandatory.

Date, Time and Location

Sep 23, 2026
5:30 PM - 8:30 PM CEST
Impact Hub Munich, Gotzinger Str. 8, 81371 München, Germany

Meetup Speakers

Online Monitoring of Mechanical Loads

Vehicle components are constantly affected by forces caused by acceleration, braking, road conditions, and driving behavior. These forces create material stresses, and repeated stress over time leads to fatigue—one of the main causes of component failure. Understanding how loads translate into stress and ultimately into damage is therefore essential.

In this talk, we present how a machine learning model can estimate mechanical loads even in vehicles without dedicated sensors for this purpose. By reconstructing hidden stresses from available vehicle data, the solution enables real-time fatigue assessment, improves component durability predictions, and supports safer, more reliable vehicle designs—without requiring additional hardware.

Speaker: Alexander Nenninger at NTT DATA

Physical Intelligence: Building the Next Gen of Robotics

Explore how physical AI is transforming robots to perceive, reason, learn, and act in the real world. Discover the Intel Robotics AI Suite, Physical AI Toolkit, and Physical AI Studio, and how developers can build, deploy, and accelerate the next generation of intelligent robots through an open community.

Speaker: Jayabalaji Sathiyamoorthi at Intel

Meetup Agenda
5:30–6:30 PM

• Networking, food, drinks, and lightning talks

Workshop Agenda
This hands-on session uses DROID, a real-world robotics dataset loaded into FiftyOne as a native multimodal MCAP recording, and YOLO11n, fine-tuned live during the session.

6:30–7:30 PM

• Welcome + framing: from raw robot logs to a trained detector
• Explore a real DROID robotics recording in FiftyOne's native multimodal MCAP viewer: camera, proprioception, and language on one synced timeline, no ROS install required
• Curate: extract and browse frames from the recording, filter and deduplicate
• Compute embeddings on the curated frames; explore via similarity search and embeddings visualization (via Nebius Serverless AI Jobs)

7:30–8:00 PM

• Auto-label: open-vocabulary detection to generate bounding boxes for the robot gripper and target objects
• Train: fine-tune a YOLO11n detector on the auto-labeled frames (via Nebius Serverless AI Jobs)
• Evaluate results and close the loop: view predictions back on the original MCAP timeline

8:00–8:30 PM

• What else Nebius offers: Token Factory walkthrough — chat/vision models, fine-tuning, credits