leveLXData by fka

🚀 Reducing Distribution Shifts in Autonomous Vehicles 🚗 Congratulations to Christopher Diehl, Peter Karkus, Sushant Veer, Marco Pavone, and Torsten Bertram from Technische Universität Dortmund, NVIDIA, and Stanford University, on their fantastic paper, "LoRD: Adapting Differentiable Driving Policies to Distribution Shifts"! The research addresses a crucial problem in autonomous driving: how to adapt self-driving models to changes in their environment—such as varying traffic rules, vehicle types, or weather conditions—without losing the knowledge they’ve already acquired. To tackle this, the authors propose a novel adaptation technique called the Low-Rank Residual Decoder (LoRD) to mitigate issues like catastrophic forgetting. To evaluate their concept, the team used leveLXData’s #exiD dataset, featuring interactive real-world highway data from Germany—perfect for testing complex self-driving models in diverse scenarios. If you're curious about the exiD dataset or how our data can support similar cutting-edge research, feel free to get in touch! ✉ #AutonomousDriving #MachineLearning #Innovation #exiD #TrajectoryData #ScenarioData #DroneData

🤝 Meet us ASAM International Conference 2024 Our colleague Thomas Keutgens will represent #leveLXData and fka GmbH at the ASAM e.V. - Association for Standardization of Automation and Measuring Systems International Conference 2024 in #Munich! This year’s event focuses on standards for ADAS / AD and the software-defined vehicle — and we’re looking forward to discuss experiences in working with some of those standards. 📌 Don’t miss our poster titled "From Real-World Data to Requirements: Scenario-based Specification for ADAS", where we showcase how high-quality, naturalistic data can help to efficiently derive requirements and accelerate ADAS development. The #ASAM Conference is a fantastic platform, and we’re looking forward to connecting with old and new partners there. If you’re attending, make sure to stop by and say hello! See you in Munich! #ASAM2024 #ADAS #AutomatedDriving #TrafficData #Simulation #Innovation #ScenarioData #OpenDRIVE #OpenSCENARIO #OpenLabel #OpenODD #DrivingBehavior #Pitch

🇨🇦 Spotlight on our #Canada Highway Data #leveLXData offers high-quality naturalistic traffic data and scenarios data from all over the world. Want to check behavioral differences for #Canadian drivers? Get some valuable insights from traffic analysis of our #Vancouver data. Here's a quick overview: - 3 locations with on- and off-ramps - +9 hours of data - +99,000 trajectories - Classes: Car, Truck, Bus, Van, Pickup, Motorcycle This dataset offers comprehensive vehicle trajectory information and can be used with our #Lanelet2 and #OpenDrive maps for deeper analysis. If you're working on #TrafficManagement, #AutonomousDriving, or #MobilityResearch, our data and methodology can help to accelerate your research and development. Reach out to us if you’re interested in learning more or have specific data requirements! #TrafficData #leveLXData #Innovation #Mobility #DroneData #ScenarioData #DrivingBehavior #DrivingData

⚡ 291,890 accidents with bodily injury in Germany (2023) 🚗 Even though especially active safety technology advances, the number of traffic accidents is still alarmingly high in Germany. In this, some roads seem to pose higher challenges for road users. For example, statistics show that Trierer Straße in #Aachen is one of the most dangerous roads in Germany with frequent accidents due to heavy traffic and complex road structures. High-quality traffic data is crucial for understanding accident patterns and making roads safer. We’ve collected comprehensive data from high-risk areas including Trierer Straße to support in-depth analysis and support in improving traffic safety measures. #leveLXData 's detailed insights into road user behavior can help cities, researchers, and technology providers to find solutions for reducing accidents and improving the overall driving experience. 🚦📊 #TrafficSafety #DataForGood #TrafficData #AccidentStatistics #leveLXData #Mobility #ScenarioData #DroneData #TrajectoryData

🚀 We're happy to support the Hi-Drive project. The European Union-funded Hi-Drive project is an ambitious initiative, aimed at advancing connected and automated driving (CAD) systems, to tackles some of the biggest challenges in autonomous vehicle development, focusing on expanding operational design domains, improving safety, and reducing reliance on human intervention. This includes: - Complex traffic including merging and exiting - Coping with adverse weather - Complex interaction with other road users including VRUs in normal traffic - Dynamic traffic rules and signs - GNSS occluded areas like tunnels It's great to help several work packages of this initiative with our drone-based naturatlistic trajectory data. Among other, our #exiD and #inD datasets are used for impact assessment, scenario and edge case detection as they cover large amounts of urban and highway scenarios. The use of these datasets by talented researchers like Marcel Sonntag, Lennart Vater, Henri Sintonen, Saeed Rahmani underscores the real-world applicability of high-quality traffic data for developing automated driving functions. One concrete example of using our data can be discovered in the paper by Marcel Sonntag, Lennart Vater, Roman Vuskov, Lutz Eckstein with the title: "Detecting Edge Cases from Trajectory Datasets Using Deep Learning Based Outlier Detection". The link to the paper can be found in the comments below 👇 🔎 Interested in learning more about how our datasets can support your projects? Reach out, and let’s explore the potential for collaboration! #HiDrive #AutomatedDriving #TrafficData #ADAS #SafetyValidation #Innovation #DroneData #ScenarioData #EdgeCase #ScenarioDetection #ImpactAssessment

🚀 Use #Lanelet2 maps to boost your applications 🌍 At #leveLXData, we support the Lanelet2 format, a high-definition map framework designed specifically for automated driving. Developed by Fabian Poggenhans, Jan-Hendrik Pauls, Johannes Janosovits, Stefan Orf, Maximilian Naumann, Florian Kuhnt and Matthias Mayr from FZI Forschungszentrum Informatik and the Karlsruher Institut für Technologie (KIT), #Lanelet2 offers powerful tools to model complex road networks. With the open-source C++ and Python library, you can: - Match our trajectory data with the provided maps for deeper insights. - Perform enhanced data analysis by leveraging lane-level geometry and map-based relationships. - Integrate the maps seamlessly into autonomous driving simulations and path planning tools. We encourage all developers and researchers to explore the Paper and the Lanelet2 GitHub repository linked in the comments below, where you'll find handy tools that enhance the possibilities of our dataset. #Lanelet2 #HDMaps #AutonomousDriving #leveLXData #OpenSource #Mobility #DroneData #TrajectoryData #ADAS

Yesterday, I had the pleasure of presenting leveLXData by fka at the International ELIV Congress's Lightning Talks, hosted by VDI e.V.! In just a few minutes, I showcased how fka GmbH is driving innovation and creating new ideas for automated driving. With leveLXData we provide high-quality naturalistic traffic and scenario data for a wide range of applications, from simulation to validation of automated driving systems. A big thank you to everyone who attended the lightning session. Let’s connect and explore new opportunities together!

🚀 Meet Us at ELIV 2024 in Bonn! We’ll be giving a Lightning Talk at VDI e.V. ELIV Congress 2024 – the world’s largest event for automotive electronics, software, and applications! This great event brings together industry leaders, decision-makers, and innovators to explore the latest trends and advancements in the automotive world. Let's meet in Bonn on Wednesday or Thursday (October, 16/17, 2024) and talk about how our data solutions and services can help drive the future of autonomous driving and advanced mobility systems. Jens Kotte and Christoph Klas will be there to discuss new ideas. We’d love to connect and explore how we can collaborate on innovative projects. See you there! #ELIV2024 #AutonomousDriving #Innovation #AutomotiveIndustry #leveLXData #Mobility

☕ Join us for the final day at Aachen Colloquium Sustainable Mobility 2024! We're excited about the plenary discussion about "Use of Artificial Intelligence in Automotive Applications" later today with: Stephan Durach (BMW Group), Ralf Herrtwich (NVIDIA) and Nikolai Ardey (Volkswagen AG). We are looking forward to connecting with you and discussing how our high-quality datasets accelerate automotive AI applications. Stop by our booth and learn:   - How our precise, real-world driving data enables safer and more efficient development & validation   - The impact of using our datasets on reducing time-to-market for ADAS and AV systems   - Discover our newest live demo, offering insights into the exciting potential of our data Looking forward to discussing how we can collaborate and propel the future of automotive AI together! 🤝

🧭 Ensuring Best-in-Class Data Accuracy At #leveLXData, we're committed to providing precise and high-quality data for your #ADAS, #AutonomousDriving, and #Traffic applications. To ensure the reliability of our data, we regularly conduct accuracy tests, keeping our system in check. This way, we can ensure that after updates our pipeline meets (and sometimes raises) the quality standards we promise. Our #fka vehicle competence and facilities certainly help us a lot for this. Here’s a quick look at our testing methodology: - Defined test setup on our test track using a dedicated test vehicle - Vehicle equipped with RTK-GPS for precise positioning - Various driving maneuvers to evaluate the pipeline - Vehicle and drone data synchronisation and comparison - Evalution of key variables, including #GlobalPosition, #Velocity, #LongitudinalAcceleration, and #Heading. Want to learn more about the results or have specific data needs? Reach out to Thomas Keutgens or Christoph Klas! #DataQuality #AccuracyMatters #Innovation #TrafficData #ScenarioData #Automotive #leveLXData