Autonomous vehicles can behave unexpectedly, as automated systems that rely on data-driven machine learning have shown to infer false predictions or misclassifications, e.g., due to stickers on traffic signs, and thus fail in some situations. In critical situations, system designs must guarantee safety and reliability. However, in non-critical situations, the possibility of failures resulting in unexpected behaviour should be considered, as they negatively impact the passenger’s user experience and acceptance. We analyse if an interactive conversational user interface can mitigate negative experiences when interacting with imperfect artificial intelligence systems. In our quantitative interactive online survey (N=113) and comparative qualitative Wizard of Oz study (N=8), users were able to interact with an autonomous SAE level 5 driving simulation. Our findings demonstrate that increased transparency improves user experience and acceptance. Furthermore, we show that additional information in failure scenarios can lead to an information dilemma and should be implemented carefully.

In a fully autonomous driving situation, passengers hand over the steering control to a highly automated system. Autonomous driving behaviour may lead to confusion and negative user experience. When establishing such new technology, the user’s acceptance and understanding are crucial factors regarding success and failure. Using a driving simulator and a mobile application, we evaluated if system transparency during and after the interaction can increase the user experience and subjective feeling of safety and control. We contribute an initial guideline for autonomous driving experience design, bringing together the areas of user experience, explainable artificial intelligence and autonomous driving. The AVAM questionnaire, UEQ-S and interviews show that explanations during or after the ride help turn a negative user experience into a neutral one, which might be due to the increased feeling of control. However, we did not detect an effect for combining explanations during and after the ride.

Within the ongoing process of defining autonomous driving solutions, experience design may represent an important interface between humans and the autonomous vehicle. This paper presents an empirical study that uses different ways of unimodal communication in autonomous driving to communicate awareness and intent of autonomous vehicles. The goal is to provide recommendations for feedback solutions within holistic autonomous driving experiences. 22 test subjects took part in four autonomous, simulated virtual reality shuttle rides and were presented with different unimodal feedback in the form of light, sound, visualisation, text and vibration. The empirical study showed that, compared to a no-feedback baseline ride, light, and visualisation were able to create a positive user experience.

During the last decades, concepts of Intelligent Transportation Systems (ITS) were continuously adapted and improved based on new insights into human travel behavior. Drivers for improvements are the quantity and quality of available mobility data, which increased significantly in recent years. Based on travel behavior, literature proposes a large number of different solutions for next step or future location prediction. However a holistic spatio-temporal prediction, which could further improve the quality of ITS, creates a more complex task. The prediction of medium-term mobility for one to seven days is challenging in particular for atypical travel behavior, since the weekdays' order delivers no reliable indication for the next day's travel behavior. With our contribution, we explore the benefits of various prediction approaches for medium-term mobility prediction and combine them dynamically to predict individual mobility behavior for a period of one week. The derived framework utilizes an exhaustive search approach to benefit from a machine learning based clustering method on location data. In conjunction with an Artificial Neural Network, the prediction framework is robust against prediction errors created by atypical behavior. With two data sets consisting of smartphone and vehicle data, we demonstrate the framework's real-world applicability. We show that clustering an individual's historical movement data can improve the prediction accuracy of different prediction methods that will be explained in detail and illustrate the interrelation of entropy and prediction accuracy.

The prediction of an individual's future locations is a significant part of scientific researches. While a variety of solutions have been investigated for the prediction of future locations, predicting departure and arrival times at predicted locations is a task with higher complexity and less attention. While the challenges of combining spatial and temporal information have been stated in various works, the proposed solutions lack accuracy and robustness. This paper proposes a simple yet effective way to predict not only an individual's future location, but also most probable departure and arrival times as well as the most probable route from origin to destination.

The upsurge of mobile devices paired with highly interactive social web applications generates enormous amounts of requests web services have to deal with. Consequently in our previous work, a novel request flow scheme with scalable components was proposed for storing interdependent, permanently updated resources in a database. The major challenge is to process dependencies in an optimal fashion while maintaining dependency constraints. In this work, three research objectives are evaluated by examining resource dependencies and their key graph measurements. An all-sources longest-path algorithm is presented for efficient processing and dependencies are analysed to find correlations between performance and graph measures. Two algorithms basing their parameters on six real-world web service structures, e.g., Facebook Graph API are developed to generate dependency graphs and a model is developed to estimate performance based on resource parameters. An evaluation of four graph series discusses performance effects of different graph structures. The results of an evaluation of 2,000 web services with over 850 thousand resources and 6 million requests indicate that resource dependency processing can be up to a factor of two faster compared to a traditional processing approach while an average model fit of 97 percent allows an accurate prediction.

Nowadays, web services have to accommodate a significant and ever-increasing number of requests due to high interactivity of current applications. Although the built-in elasticity offered by a cloud can mitigate this challenge, it is highly desirable that applications can be built in a scalable fashion. State-of-the-art Web Application Frameworks (WAFs) focus on the creation of application logic and do not offer integrated cloud scaling concepts. As the creation of such scaling systems is very complex, we proposed in our recent work the concept of Web Scaling Frameworks (WSFs) in order to offload scaling to another layer of abstraction. In this work, a detailed design for WSFs including necessary modules, interfaces and components is presented. A mathematical model used for performance rating is evaluated and enhanced on a computing cluster of 42 machines. Traffic traces from over 25 million real-world applications are analysed and evaluated on the cluster to compare the WSF performance with a traditional scaling approach. The results show that the application of WSFs can substantially reduce the number of total machines needed for three representative real-world applications-a social network, a trip planner and the FIFA World Cup 98 website-by 32, 63 and 92 percent, respectively.

Autonomous cars are on the horizon, meaning passengers will no longer have to focus on driving leaving them with extra time for other activities, or engagements. However, research has focused primarily on safety related aspects of autonomous driving, overlooking the need to design for this new free time. This raises the question, how do we design new interactive experiences for the future of autonomous cars? In this paper, we present a design process derived from our research-through-design approach to explore possible everyday use cases of autonomous driving from an experience-focused perspective. We report details of the four methods that constituted, and influenced our design process and led to the creation of AutoGym, an exertion interface with context-based interactions suitable for future car-based commuting. The contribution is twofold: Foremost, our design process suggests guidelines on how to design and simulate future use cases of what we assume will constitute the autonomous driving experience. Secondly, we aim to inspire automotive user experience designers to pursue a context-based design approach by leveraging situational features which support experiences that are tailored and unique to autonomous driving.

The social web and huge growth of mobile smart devices dramatically increases the performance requirements for web services. State-of-the-art Web Application Frameworks (WAFs) do not offer complete scaling concepts with automatic resource-provisioning, elastic caching or guaranteed maximum response times. These functionalities, however, are supported by cloud computing and needed to scale an application to its demands. Components like proxies, load-balancers, distributed caches, queuing and messaging systems have been around for a long time and in each field relevant research exists. Nevertheless, to create a scalable web service it is seldom enough to deploy only one component. In this work we propose to combine those complementary components to a predictable, composed system. The proposed solution introduces a novel class of web frameworks called Web Scaling Frameworks (WSFs) that take over the scaling. The proposed mathematical model allows a universally applicable prediction of performance in the single-machine- and multi-machine scope. A prototypical implementation is created to empirically validate the mathematical model and demonstrates both the feasibility and increase of performance of a WSF. The results show that the application of a WSF can triple the requests handling capability of a single machine and additionally reduce the number of total machines by 44%.