Adaptive environment-dependent localization of autonomous vehicles with methods of artificial intelligence


The project aims to create the foundation towards new AI research groups composed of researches at the Otto-von-Guericke University and the ifak e.V. focusing on developing, extending and applying AI-Methods for industrial scenarios. The two junior research groups at Otto-von-Guericke University and ifak are headed by Dr. Christoph Steup and Dr. Maxim Nesterov. Furthermore, the project serves to establish connections between the current AI research at ifak e.V. and Otto-von-Guericke University with business entities.

As part of the project, the cooperation partners will address a highly relevant problem of autonomous mobility: sensitivity of driving operation to weather conditions and failure of external localization by, for example: Satellite navigation systems and cellular connectivity. The mentioned effects currently lead to a disruption of the autonomous operation.


The AULA-KI project aims at providing localization information for autonomous vehicles anytime and anywhere. For this purpose, both partners will jointly develop methods to detect, evaluate and mitigate the effects of weather conditions on typical sensors of autonomous vehicles (camera, LiDAR, odometry). In particular, existing artificial intelligence algorithms will be used and further developed for model building, but also for optimizing the output of the sensors. Demonstration and validation of the developed AI methods and localization enhancements will be performed with the help of an autonomous passenger shuttle on a dedicated test site. The project results lay a foundation for reliable, autonomous and sustainable mobility of tomorrow.


Technical objectives:

  • Quality assessment of sensor data and quality metrics: Analyze vehicle localization and related sensor data. Determine the specific challenges of accurate localization and develop an AI-based evaluation algorithm to assess the current weather situation and localization quality of the vehicle.
  • Factoring out weather effects: Develop AI-powered algorithms that allow weather events to be "hidden" from the actual sensor data, thus avoiding delocalization.
  • Localization supported by swarm intelligence and infrastructure: For situations in which "factoring out" weather events is not possible, or difficult weather conditions are supplemented by additional, temporarily occurring deteriorations in localization quality (mobile radio failure or GPS malfunctions), localization is to be ensured by additional data sources: Infrastructure communication, swarm intelligence, data fusion.


Associated partners:
- EasyMile GmbH
- RBO Regionalbus Ostbayern GmbH
- Gestalt Robotics GmbH
- Leipziger Verkehrsbetriebe GmbH