Integration of real-time situational awareness in real-life automated vehicles

Problem Context

Self-driving vehicles could eventually make our mobility and logistics more efficient, cleaner and safer in the future. But right now, decision-making is one of the most challenging tasks for an automated vehicle. State-of-the-art solutions are either demonstrably safe, but unable to scale up to complex situations such as driving very slowly or stopping for swirling leaves. Alternatively, the solutions are able to handle complex situations well, but their safety is difficult to assess. This is of no benefit to safety or efficiency and causes many so-called 'false positive stops'.


At TNO, we are working hard to develop AI algorithms to teach self-driving vehicles to respond like a human driver. This demo shows how this Hybrid-AI approach allows a vehicle to handle complex traffic situations and take into account the risks associated with actions. At one point, the lane is blocked by an obstacle. The vehicle must then decide whether it is safe to plan and execute a trajectory around the obstacle, or wait behind it.

What makes this unique?

Joëlle van den Broek, Principal Consultant Smart and Safe Mobility: “The vehicle uses 'context-dependent behaviour' AI to plan its route. This technology is still in the stage of knowledge development, but offers very interesting application possibilities for numerous automated driving applications, such as shuttle services or Connected Automated Transport."


  • Joëlle van den Broek, Principal Consultant Smart and Safe Mobility, e-mail: