Explainable Online Lane Change Predictions on a Digital Twin with a Layer Normalized LSTM and Layer-wise Relevance Propagation

Artificial Intelligence and Digital Twins play an integral role in driving innovation the domain of intelligent driving. Long short-term memory (LSTM) is a leading driver field lane change prediction for manoeuvre anticipation. However, decision-making process such models complex non-transparent, hence reducing trustworthiness smart solution. This work presents innovative approach technical implementation explaining predictions layer normalized LSTMs using Layer-wise Relevance Propagation (LRP). The core includes consuming live data from digital twin on German highway, explanations changes by extending LRP to LSTMs, interface communicating human user. We aim demonstrate faithful, understandable, adaptable increase adoption AI systems that involve humans. Our research also emphases explainability state-of-the-art performance ML anticipation go hand without negatively affecting predictive effectiveness.