This videos demonstrate the CARLA simulation validation of the CARE-Drive configuration (M* = gpt-4.1, T* = Tree-of-Thought) for evaluating reason-responsive decision-making in automated driving. The
Predictions on the actions-features matrix. The indicators (10) are computed for different levels of information used as “training set”: more precisely, the different values of T* correspond to 75%, 5