From June 4th to 6th, 2025, our partner, the Technical University of Crete (TUC), ensured the scientific excellence of the iDriving project by presenting a technical paper at the IEEE International Conference in Electronic Engineering & Information Technology.
The paper, titled “Addressing Reward Hacking in Autonomous Navigation using a Soft Actor Critic approach,” offers a deep dive into the advanced Artificial Intelligence (AI) methodologies being developed within the project. Specifically, it explores how to make autonomous navigation safer and more robust against common AI pitfalls like “reward hacking.”
This presentation served as a critical milestone for scientific dissemination. By engaging with the global engineering community, TUC subjected the project’s algorithms to rigorous peer review. This process is essential for validating the Soft Actor-Critic (SAC) approach that iDriving employs in its real-world pilots.
Furthermore, the conference provided a platform for exchanging best practices on reinforcement learning in autonomous systems. The feedback gathered from external experts ensures that iDriving’s navigation solutions remain at the cutting edge of technology.
