Join us online for the 7th International Workshop on Naturalistic Driving Data Analytics (NDDA) on Friday 30th of October, organized by Pujitha Gunaratne Phd, Principal ScientistToyota Motor North AmericaAnn Arbor, MI, USA and Prof, Kemal Ure , Director of AI at Eatron, as a part of IEEE Intelligent Vehicles Symposium – IV 2020. This year we have an amazing array of invited speakers and papers, make sure that you check out the program from the link below.

As a part of the workshop, Eatron Technologies Ltd will be also showcasing our latest .

Safe lane changing is a crucial part of highway driving since it requires anticipation of the traffic flow and interaction with the other drivers. The automated lane change functionality is gaining popularity as a common feature in the L2+ ADAS systems. That being said, designing automated lane change functionality that is safe, accepted by the human driver and scalable to the many different conditions, is still an open challenge. In order to overcome this challenge, we propose a deep reinforcement learning based automated lane change algorithm that utilizes simulation-to-real (sim2real) and real-to-simulation (real2sim) data transfers. We demonstrate that sim2real transfer performs better when the simulation scenarios are sampled from real world data that contains naturalistic driving behaviors and traffic flow dynamics. In addition, we demonstrate our evaluation procedures for comparing reinforcement learning based lane change decisions to natural driver decisions.