Pedestrian Crossing Prediction (Explainable AI)

Short-horizon prediction of pedestrian crossing-corridor entry using real-world trajectory data from Viscando (Gothenburg).

Master's degree project at Skövde comparing logistic-regression baselines, an interaction-aware MLP, and a compact GRU sequence model across 2/3/5-second horizons. Emphasis on interpretable feature engineering and permutation-importance analysis for ADAS and road-safety use.