Fusion of uncertain estimates for the position and spatial extent of a dynamic object

• Applicant: Professor Dr.-Ing. Marcus Baum
• Subject Area: Automation, Mechatronics, Control Systems, Intelligent Technical Systems, Robotics
• Term: since 2026
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 559624998

The sensor-based perception of the environment is a central task of autonomous systems such as self-driving vehicles. In this context, this project is concerned with the determination of the position and spatial extent of a dynamic object. Due to the perspective, the measurement principle and measurement noise, a single sensor typically only obtains an incomplete and erroneous picture of the spatial extent. For this reason, this project deals with the question of how information from different (heterogeneous) sensors about an extended object can be fused. A central challenge is the systematic incorporation of uncertainties about the spatial extent. The key idea of this project is to develop Bayesian estimators that employ a distance measure on geometric figures as a cost function. By this means, the geometric properties of the spatial extent can be systematically incorporated in the estimator. In a first step, the project investigates suitable distance measures and representations for spatially extended objects. Next, models for the uncertainty of the extent estimate and methods for calculating the expected value are investigated. Then, fusion methods for the combination of different estimates for a spatially extended object are developed. First, the case of independent estimates is considered, and second, the case of unknown dependencies, which is in particular relevant for distributed sensor networks, is treated. An evaluation of the fusion methods is performed with the help of simulated and publicly available real data in the context of traffic surveillance.

• DFG Programme: Research Grants