Simulation of the Effects of Different AI Distributions on Urban Mobility Systems

Context

The “Future Habitat Mobility” research project investigates how AI-based functionalities can be efficiently and purposefully distributed among vehicles, infrastructure, and the cloud in future mobility systems. The focus is on determining which tasks should be performed at which system level to optimally meet requirements regarding performance, safety, scalability, and sustainability. The emphasis is on developing and evaluating future-proof system architectures for connected and autonomous mobility solutions. Both technical and systemic aspects are taken into account, such as latency, data availability, robustness, energy efficiency, communication overhead, and security and data protection requirements. A central component of the project is the model- and simulation-based analysis of various distribution strategies—ranging from purely vehicle-based approaches to hybrid architectures and cloud-centric concepts. These strategies are examined and compared using realistic use cases from the Division of intelligent mobility in order to draw well-founded conclusions about their potential, limitations, and application scenarios.

Targets

• Analysis and comparison of different strategies for distributing AI functionalities among vehicles, infrastructure, and the cloud

• Development and evaluation of future-proof system architectures for connected and autonomous mobility systems

• Modeling and simulation-based investigation of realistic mobility use cases

• Evaluation of the strategies in terms of performance, safety, latency, robustness, energy efficiency, and sustainability

• Identification of suitable architectures and distribution approaches for different application and operational scenarios

Requirements

• Knowledge of Python
• Experience with the CARLA simulation environment is a plus