Our traffic agent models

Once autonomous vehicles become a mainstream reality, they will share the road with us humans. Mixed traffic involving both autonomous vehicles and human road users pose a major challenge for our future mobility. Extensive testing in simulation is the solution to ensure road safety in mixed traffic scenarios. To harness the full potential of simulation, the simulated world must mimic reality as closely as possible.

The most challenging aspect is to realistically simulate human road users and their seemingly unpredictable behavior.
cogniBIT fills the simulation with life-like road users by providing traffic agent models which perceive, think and act like human road users and can be integrated in any simulation platform.

Closed-loop simulation

cogniBIT’s algorithms run alongside the simulation core. Instead of using pre-generated, open-loop trajectories, our traffic agent models realistically interact with the vehicle under test in real time. The interaction between the automated driving function and human road users can thus be optimized.


For the development of autonomous driving, it is essential to reconstruct critical scenarios in simulation. Our white-box approach provides transparent and traceable behavior of the simulated road users. During a virtual analysis of critical scenarios, accidents can be traced back to different causes such as distraction, lack of attention, or distance estimation errors. In this way, accidents caused by other drivers’ misbehaviour can easily be distinguished from accidents caused by the autonomous vehicle under test.


No two drivers are the same. Human behavior in traffic is influenced by a large variety of factors, such as age, driving experience, and emotions. In our model-based approach, all of these factors can be combined and adapted for each simulated individual driver to bring the entire continuum of human driving styles into simulation.
From the defensive, safety-conscious elderly driver to the young hot shot, cogniBIT is able to simulate the behavior of the entire range of human drivers.

Critical Scenarios

Safely mastering critical scenarios such as accidents and near-accidents represents the benchmark for the general acceptance of autonomous vehicles. Such scenarios are mostly rooted in human error. By simulating human behavior and its limitations, our systematic variations of scenarios automatically generate plausible and life-like traffic flows. Thus, cogniBIT provides developers of ADAS and autonomous driving technology with the tools to efficiently test critical scenarios.

cogniBIT enables ...

Autonomous driving companies

to develop safe vehicles for all traffic situations.

Regulatory agencies

to assess road safety of new mobility technologies.

Urban planners

to make road networks ready for mixed autonomous and human traffic.