The ELLIS Summer School on Autonomous Driving brought together researchers, PhD students and industry experts in Barcelona to discuss current and future developments in autonomous driving. The summer school was organised by the Computer Vision Center, the ELLIS Unit Barcelona and Valeo. Around 70 participants attended the event.
Among them was Ramakrishnan Subramanian, research associate at inIT. In the Intelligent Systems research group headed by Prof. Dr. Ulrich Büker, he is conducting his PhD research on monitoring the Operational Design Domain of autonomous vehicles.
ODD monitoring for autonomous vehicles
During a poster session, Ramakrishnan Subramanian presented his PhD research entitled “Operational Design Domain Monitoring System for Autonomous Vehicles” and discussed his approach with experts, researchers and PhD students from different countries.
“The exchange and the constructive feedback were particularly valuable to me and gave me new perspectives for the further development of my PhD research,” says Ramakrishnan Subramanian.
From end-to-end driving to world models
The summer school programme covered a range of current research directions in autonomous driving. These included end-to-end driving, Vision-Language Models, Vision-Language-Action models, world models, dataset preparation and curation, as well as questions of safety and robustness in autonomous systems.
Speakers included José M. Álvarez, Director of AI Research at NVIDIA, and Germán Ros, Director of Simulation at NVIDIA. Holger Caesar from TU Delft spoke about dataset preparation and curation. Other contributions focused on Vision-Language-Action models at Bosch, world models at Wayve and work on GPU compilers at Google.
A recurring topic was the transition from classical modular architectures towards more end-to-end-based approaches. The discussions also addressed how new AI models can deal with complex, rare and unexpected traffic situations. These include so-called black swan scenarios, which remain a particular challenge for autonomous vehicles.
Exchange with the international research community
In addition to the talks and poster sessions, Ramakrishnan Subramanian used the summer school to exchange ideas with current and former PhD students. This gave him insights into different research questions, methods and experiences.
The exchange also showed how research groups, universities and companies around the world are addressing current challenges in autonomous driving. For his own PhD research, Ramakrishnan Subramanian gained new perspectives in particular on AI-based approaches for autonomous driving systems.
The summer school also created new international contacts and further opportunities for exchange between research at inIT and the international autonomous-driving research community.