AI-supported location and route prediction through multimodal analysis of GPS and voice data
The aim of the AURORA project is to develop an AI-based location and route prediction system that combines GPS data with contextual information from voice recordings in vehicles. Machine learning methods are used to analyze movement patterns and predict future trajectories. In addition, speech recognition, speaker identification and semantic analysis are used to extract relevant contextual information. The fusion of language of instruction and movement data increases the accuracy of destination and route predictions, even with short movement histories. The system works adaptively in dynamic traffic scenarios. By optimizing routes, the solution contributes to more resource-efficient and environmentally friendly mobility.
Scientific tasks:
- Development of ML- and DL-based methods for location and route prediction based on historical GPS data
- Development of methods for semantic context analysis (e.g. Named Entity Recognition) from language data
- Integration of speech context and movement data to improve prediction quality
- Investigation of multimodal learning approaches for the fusion of heterogeneous data sources
- Analysis of the prediction accuracy, robustness and trustworthiness of AI models
Project manager:
Project staff:
