Situation Analysis and Planning

The environmental perception optimised through data fusion enables the project members to devote special attention to issues of interaction and co-operation. This concerns the different decisions that an automated vehicle has to make in urban traffic – for example, to interrupt merging if other road users do not provide a clear space. At the same time, this phase of the project also aims to enhance the predictive, i.e. anticipatory, capabilities of automated vehicles: After all, it is crucial to constantly “bear in mind” all of the possible courses of action of all parties involved. To do this, researchers make extensive use of data-driven models, which can also be used to plan driving manoeuvres – even in exceptional circumstances. For example, when necessary, an autonomous vehicle must be able to drive over a restricted area if an ambulance is approaching from behind (a temporary but legally compliant violation).

Main Topics

 Interaction and cooperation

  • Data-driven interaction models; modelling of implicit / explicit cooperation
  • Consideration of dynamic / static context information; willingness to co-operate
  • Analysing and evaluating traffic situations and cooperation scenarios

 Prediction

  • Data-driven and multimodal prediction of road users, taking into account static context and the learned interaction and cooperation models between road users
  • Evaluation of prediction precession; self-assessment of current system capabilities

 Behaviour and manoeuvre planning

  • Planning of the ego trajectory and speed
  • Combination of model-based planning processes with data-driven AI; protection by defining a safety area