Zenseact Open Dataset updated with radar data, creating opportunities for advanced research to make self-driving technology safer.
Gothenburg, 20 January 2025. Following the release of the Zenseact Open Dataset, Zenseact researchers are elevating the dataset by adding radar data. Radar is crucial to self-driving technology because it helps the car detect its surroundings even in low light or bad weather conditions. It complements light-sensitive sensors like lidar and cameras by determining the distance and speed of objects.
Automakers have been incorporating radar into their software for quite some time. However, integrating radar into an increasingly AI-based system such as Zenseact’s is a recent development.
“If we want automation to be reliably and consistently available, particularly under various weather and lighting conditions, we need a multimodal setup where the different sensors complement each other. Adding radar data to our already open dataset is a significant enhancement,” says Maryam Fatemi, Research Lead at Zenseact.
Boosting product development and research
Multimodal perception systems rely on a subset of or all three sensors: camera, lidar, and radar. Most current multimodal solutions that use AI, i.e., machine learning, have only been tested on camera and lidar. Adding radar to these solutions and devising new ones incorporating radar, camera, and lidar – in other words, fusing the different modalities – builds a more robust understanding of the surrounding environment. This is vital for developing vehicles with higher levels of autonomy. Importantly, radar makes all automated driving systems safer, and it’s even more critical for cars not equipped with lidar.
Moreover, by adding radar data – with generous permission from Continental, Zenseact was permitted to publish the data collected using Continental’s Front Looking Radar ARS513 – Zenseact enables broader research investigations, making the dataset closer to realistic autonomous driving setups. So, even if the driving scenarios or scenes haven’t changed (the radar data was collected simultaneously with the rest of the ZOD data), the new modality enhances the dataset’s usefulness for the community.
“Continental is excited to contribute to The Zenseact Open Dataset and to support Zenseact’s initiative to enable advancements in autonomous driving technology. The addition of Continental’s radar data to the dataset will address a crucial gap in sensor data and empowers researchers with richer information. Radar is an essential sensor in next-generation automotive solutions, and this addition will boost the development of safer Assisted and Automated Driving functions, ” says Dr. Markus Jüngst, Head of Concept Development Radar at Continental.
What is Zenseact Open Dataset?
ZOD is a large multimodal autonomous driving dataset created by Zenseact researchers and developers. It was collected over two years in 14 European countries using a fleet of cars equipped with a full sensor suite. ZOD aims to advance AD technology by providing industry peers and researchers with diverse, anonymized traffic data with unparalleled sensor range and resolution. ZOD comprises several hours of data from various high-range and resolution sensors, including an 8 MP camera, lidar, GNSS/IMU, radar, and vehicle data, making it an extensive multimodal dataset.
“ZOD provides various driving scenarios captured in multiple conditions, from Swedish winter roads, rainy Polish highways, and sunny Italian suburban streets. It also contains a comprehensive set of high-quality, detailed annotations for several perception tasks. By adding radar as a new modality, ZOD becomes even more versatile and valuable, enabling researchers to tackle complex challenges in autonomous driving, such as perception in adverse weather, enhanced depth estimation, and reliable object tracking. This addition enriches the dataset, allowing for more robust, real-world applications and advancing the development of autonomous vehicle technology, “says Mina Alibeigi, research lead at Zenseact and leader of The Zenseact Open Data set.
The dataset consists of three subsets: Frames, Sequences, and Drives, designed to encompass data diversity and support for spatiotemporal learning, sensor fusion, localization, and mapping. Together with the data, we have developed a software development kit containing tutorials, downloading functionality, and a dataset API for easy access to the data. The development kit is available on Github.
What is the availability of public datasets with good radar data?
While large-scale multimodal AD datasets are already available, only a few are comprehensive enough to reflect the diverse range of driving scenarios that self-driving vehicles will encounter in real life. There are several public datasets containing radar data, but combining the following factors makes Zenseact Open Dataset stand out:
- The quality of the radar data and diversity of the driving scenarios in terms of location, weather conditions, and driving scenarios
- Presence of long-range radar data
- Permissive license of the dataset
Who can use the dataset?
Anyone interested can request and be granted access. The dataset’s permissive license allows both research and commercial use (subject to the license terms) and sharing and adapting permits, providing an opportunity for startups and other commercial entities to leverage the dataset for their projects. For more information about Zenseact Open Dataset, visit https://zod.zenseact.com/
About Zenseact
Zenseact is a global AI company that develops car safety software for autonomous driving (AD) and advanced driver assistance systems (ADAS). The company has around 800 employees and works with Volvo Car Group and Polestar to enable the next big step in automotive safety. With a combination of rule-based code and deep learning algorithms, the company’s technology covers every aspect of automation, from sensor fusion, computer vision, and object detection to localization and actuation. The teams are based in Gothenburg, Lund, Munich, and Shanghai.
Editorial contact
Veronika Nihlén, Head of Communications
Email: veronika.nihlen@zenseact.com
Telephone: +46721799103
