CASE 2024 Summer school on Autonomous Systems for Smart Cities

9:00-10:30 | Birgit Vogel-Heuser

Digital Twins to Improve Engineering and Operation of Machines and Plants for Increased Autonomy

Digital Twins in automated Production Systems shall support longevity of such systems as well as adaptability and thereby enable Industry 5.0 and autonomy in the future. The talk will introduce the specific requirements, challenges and potential benefit of Digital Twins for such long living aPS as well as the technologies used. Three main aspects are discussed; model driven engineering for traceability and maintenance, (De-)Composability of control software to improve adaptability and autonomy and evolvability of DTs by learning during operation. Application examples like the extended Pick and Place Unit, the MyYoghurt plant and the Bavarian KI. Fabrik illustrate the approach

10:30-12:00 | Fabio Bonsignorio

Reproducible and Measurable research in AI and Robotics for Smart Cities

We will highlight the importance of a wider adoption of reproducibility and benchmarking methods to accelerate research and ease the technology transfer of research results in AI and Robotics. We will discuss the importance of reproducible research for dependability in smart city applications. We will illustrate the epistemological and theoretical basis of current methods and describe the best practices. We will have a look at closer and less close future developments.

12:00-13:30 | Bengt Lennartson

Robust and Data-efficient Reinforcement Learning and Fast Optimal Planning

Model-free reinforcement learning (MFRL) is very popular, but practical experience shows that MFRL both requires a lot of data and is very sensitive to high frequency resonances. Furthermore, the reward shaping is both critical and time consuming. In this talk, these problems are illustrated, and a couple of simple and efficient strategies are proposed. The strategies are applied to energy optimization of moving devices, including eco-driving of electric cars and trucks. It is also shown how constraint programming (CP) combined with a satisfiability (SAT) solver can be used for planning, based on Petri nets. Optimal plans can then be generated 100-1000 times faster compared to standard strategies.

14:30-16:00 | MengChu Zhou

Autonomous Vehicle Group: Formation, Cooperation and Evolution in Internet of Vehicles

Vehicle groups that are composed of autonomous vehicles can increase the perception range of vehicles, and their dynamic evolution can provide guidance for the effective operation of autonomous vehicles. Most existing studies on vehicle group formation neither propose a standard vehicle group model, nor consider vehicle mobility and dynamic topology of vehicle groups. Instead, they focus on detecting dynamic evolution without predicting it. This talk presents the dynamic formation, cooperation, evolution and prediction for autonomous vehicle groups, thus benefiting the effective operation of smart cities.

16:00 – 17:30 | Maria Pia Fanti

New Approaches for Managing Urban Mobility Involving Cooperative, Connected and Automated Vehicles

The number of cooperative, connected and automated vehicles (CCAVs) in urban areas will gradually increase soon. As a consequence, mixed traffic made of both regular human driven and CCAVs will likely be a typical scenario over the next few years. The short course will present some approaches for cooperative, connected and automated mobility. In particular, some strategies based on Artificial Intelligence tools and traffic simulation will be presented for collision detection and avoidance at intersections, route planner in the urban areas and car pooling services.