Vision geometry and deep neural networks – how to verify deep learning detection with vision geometric principles

Insights of AI to increase testing coverage of AVs – how to validate data using qualitative approaches

Reinforcement learning for vehicular path planning – how to make use of deep learning for AVs

Full stack software suites for AI – how to develop hardware-agnostic and scalable solutions

AVs and simulation – how to enable deep learning with synthetic data

Ensuring safety of self-driving cars – how to prove that AI systems and vehicles as a whole fulfill all safety goals

Behavior learning for AVs – how cars can learn from real world human driving behavior

Self-supervised learning for AVs – how to collect and label data at scale

Data processing and AI – how to develop software architectures and overcome hardware challenges

Path planning and object recognition with AI – how to include deep learning in the mix of computer vision approaches

Interpretable learning for AVs – how to use visual explanations that causally influence CNN output

Perception-driven machine learning for AD/ADAS systems – developing benchmarks for camera- and lidar-related perception

Next-generation radar – how to include intelligent radars into AV sensor suits

The AV electronics domain – how to move from proof of concept to purpose-build production-ready L4/L5 autonomous vehicle electronics

Automotive camera technology and computer vision algorithms – how to turn cameras into primary sensors for object recognition and classification, localization, decision making, trajectory planning, and vehicle control

Open and modular verification and validation – how to manage rapidly changing test requirements

Sensor fusion – how dot properly weight sensors at run-time

Lane detection and tracking – how to achieve robustness and accuracy in difficult scenarios

Perception stress testing – how to automatically identify where perception is brittle using unlabelled data