Teaching Cars to Predict Humans: A PhD Journey in AI for Autonomous Driving

Lina ACHAJI

LinkedIn, ThesesFr

Merci beaucoup pour votre témoignage

Could you tell us about your background before starting your PhD?

After completing my engineering studies in Lebanon, I specialized through a master’s degree in Artificial Intelligence, and Robotics. Throughout my studies, I was deeply fascinated by the intersection between software intelligence and physical systems. Autonomous vehicles represent one of the most compelling manifestations of this combination: a machine capable of perceiving its environment, reasoning, interacting with humans, and driving itself.

During the final years of my studies, I began reading extensively about autonomous vehicles, particularly the DARPA Grand Challenges and the scientific difficulties behind them. A pivotal moment in my scientific journey came when I learned that Sebastian Thrun’s team won the competition not through mechanical ingenuity alone, but by enabling the vehicle to learn, adapt, and generalize autonomously. This realization fundamentally shaped my ambitions and motivated me to pursue research in artificial intelligence applied to autonomous driving.

Why did you choose to pursue a CIFRE PhD?

After successfully completing my internship at Inria in Nancy, where I worked on topics closely related to human and robot localization, I was selected to pursue a CIFRE PhD with Stellantis. Stellantis is one of the pioneering industrial actors in France conducting advanced research on autonomous driving, with the goal of enabling vehicles to navigate safely and autonomously from point A to point B.

The collaboration between Inria and Stellantis greatly facilitated my PhD journey. Inria provided a strong scientific environment, access to state-of-the-art research resources, and the academic framework needed to establish solid research foundations. Stellantis, on the other hand, offered engineering expertise, real-world data, and the tools required to translate research ideas into practical solutions.

Working within an industrial setting also created a valuable feedback loop, helping me identify which research problems are truly impactful and relevant for real-world applications.

Can you summarize your PhD work in a few words?

My PhD focused on understanding and predicting human behavior around autonomous vehicles, enabling the vehicle to anticipate future situations and plan its actions accordingly. This is a highly multidisciplinary topic, involving elements of robotics, psychology, artificial intelligence, and human–machine interaction.

The AI component was particularly exciting. I started my PhD in 2020, at a time when attention-based models and Transformers were rapidly gaining momentum. This allowed me to study these architectures in depth and apply them to behavior prediction and multi-agent interaction modeling.

The expertise I developed during this period later proved extremely valuable, especially with the rapid rise of large language models and foundation models in the years that followed.

What advice would you give to students considering a PhD?

Be curious and genuinely fall in love with what you are working on. A PhD is not just three or four years of your life—it can shape your entire future if approached with the right mindset.

This is particularly true in artificial intelligence, where researchers with strong scientific training and critical thinking skills are highly sought after in industry.

By the end of your PhD, aim to become a scientist who can question assumptions, reason rigorously, and adapt to rapidly evolving fields.

Thanks to my PhD work - followed by two additional years of research at Stellantis - I am now working as an Applied Scientist at Amazon, where I continue to apply and extend my AI expertise beyond the scope of my original thesis topic.