Publikationen

Plank, M., Lemardelé, C., Assmann, T., Zug, S.
Journal of Urban Mobility, Volume 2, December 2022.
DOI: 10.1016/j.urbmob.2022.100036

Abstract: Unlike autonomous car applications, the operational area of urban service autonomous robots like autonomous delivery robots (ADRs) is not clearly defined at the moment. Due to large variations in the different robot designs, specific local infrastructure and regulation, assessing the feasibility of different operational scenarios is difficult. This paper presents a prototype evaluation methodology based on Open Street Map data for the assessment of ADR deployments considering one-to-many delivery schemes. Four different robot configurations and potential operational specifications are modeled and evaluated in a sample of German cities. The bandwidth of considered robot types ranges from large ADRs operating on roadways down to small size systems operating on sidewalks. The performance of the first category is limited by the reduced accessibility in areas with higher traffic. On the contrary, small ADRs present a higher detour time but increased accessibility. The evaluated operational scenarios show very diverse performance depending on the considered metrics and cities. For all the metrics considered in the paper, sidewalk ADRs show poor performance when compared to other potential ADR deployments.

Zug S., Seyffer N., Plank M.;, Pfleging B., Schrödel F., Siebert F. W.
Conference: 2023 IEEE 28th International Conference on Emerging Technologies and Factory Automation (ETFA)
DOI: 10.1109/ETFA54631.2023.10275428

Abstract: There is great application potential for mobile robots that move along sidewalks, footpaths and cycle paths. At the same time, the legal framework for their operation is currently lacking, so that a realistic evaluation of the suitable characteristics of a robot, its behavior, and the resulting acceptance by the population is still pending. The currently running project RoboTraces aims to fill this gap and classifies the behavior of people during encounters with autonomous systems in everyday life. For this purpose, a multi-month study is currently being conducted in which a robot moves through Freiberg, a medium-size city with 40,000 inhabitants in Central Europe. We are recording the reactions of passers-by, classifying them according to the context of the situation and assessing their significance for the planning of movement corridors for small mobile systems.

Plank M., Jäger G., Zug S.
Conference: 2023 IEEE International Symposium on Robotic and Sensors Environments (ROSE)
DOI: 10.1109/ROSE60297.2023.10410711

Abstract: The configuration of autonomous cars (maximum dimension, features, behavior rules) and the operation area are well-defined. Hence, for a dedicated road, the mission planner can assume that the system can move on it, at least in principle. For smaller robots operating on footpaths and cycle tracks this prior knowledge does not exist. From an infrastructure perspective, for example, the widths of pedestrian and bicycle vary widely and integrate different types of obstacles. On the other hand, the robot itself (size, kinematics, sensory equipment, etc.) is not standardized. For mapping infrastructure and robot's properties, a time-consuming manual analysis and experiments with the actual robot in the intended area of operation is necessary. This approach is unrealistic for the real-world, wide spread application of mobile robots - the data for the navigability analysis must be collected using a different strategy.

In order to automate this process, the paper examines the state of the art in the collection of environmental data. It extracts the parameters relevant for the operation of a robotic system from previous work, compares them with established data collections / services and identifies their limits when analyzing possible application areas. Building on this, the paper evaluates 150 projects that aim to close these gaps with different concepts. The article structures and classifies the projects and analyses their focal points. We can conclude that only a multi-model approach is able to address the complexity of acquiring all necessary environmental data with a high coverage. Additionally, with that analysis, we provide methodological support to plan future robotic applications.

Zug S., Jäger G., Syffer N., Plank M., Licht G., Siebert F. W.
Conference: 2024 IEEE International Symposium on Robotic and Sensors Environments (ROSE)
DOI: 10.1109/ROSE62198.2024.10590842

Abstract: This study identifies a gap in data-driven approaches to robot-centric pedestrian interactions and proposes a corresponding pipeline. The pipeline utilizes unsupervised learning techniques to identify patterns in interaction data of urban environments, specifically focusing on conflict scenarios. Analyzed features include the robot's and pedestrian's speed and contextual parameters such as proximity to intersections. They are extracted and reduced in dimensionality using Principal Component Analysis (PCA). Finally, K-means clustering is employed to uncover underlying patterns in the interaction data. A use case application of the pipeline is presented, utilizing real-world robot mission data from a mid-sized German city. The results indicate the need for enriching interaction representations with contextual information to enable fine-grained analysis and reasoning. Nevertheless, they also highlight the need for expanding the data set and incorporating additional contextual factors to enhance the robots situational awareness and interaction quality.