Kai von Luck

Semiautomatic and user-centered orientation of digital artifacts on multi-touch tabletops

The orientation of objects on tables is of fundamental importance for the coordination, communication and proper understanding of content in group work. Similarly, the roles of orientation have to be taken into account when implementing software for multi-touch tabletops. This paper describes a combined approach to help with the orientation of artifacts, composed of both automatic and manual orientation methods. Using a custom test application, this study investigates the effects of automatic orientation of artifacts towards users.

Physiological data analysis for an emotional provoking exergame

In this work we enhance our previously developed analysis method of provoked emotions in facial expressions through the analysis of physiological data. The presented work describes the integration of electrodermal activity, respiration and temperature sensors to enhance our exergaming system for emotional provocation. The combined analysis of facial expressions and physiological data is designed to evaluate physical and cognitive stress as well as emotional reactions. The experimental setup combines a cycling game controller with a 3D virtual cycling game to provoke emotions. A designed data recording framework collects frontal videos and physiological data as well as game and controller events. In this work, we found evidence that physiological data analysis enhances the previously developed analysis method. The system is able to evaluate individual differences of an entertaining and balanced workout program.

Towards More Robust Automatic Facial Expression Recognition in Smart Environments

In this paper, we provide insights towards achieving more robust automatic facial expression recognition in smart environments based on our benchmark with three labeled facial expression databases. These databases are selected to test for desktop, 3D and smart environment application scenarios. This work is meant to provide a neutral comparison and guidelines for developers and researchers interested to integrate facial emotion recognition technologies in their applications, understand its limitations and adaptation as well as enhancement strategies. We also introduce and compare three different metrics for finding the primary expression in a time window of a displayed emotion. In addition, we outline facial emotion recognition limitations and enhancements for smart environments and non-frontal setups. By providing our comparison and enhancements we hope to build a bridge from affective computing research and solution providers to application developers that like to enhance new applications by including emotion based user modeling.

Ambient Surfaces: Interactive Displays in the Informative Workspace of Co-located Scrum Teams

This paper contributes long-term findings on a custom ambient display solution (Ambient Surfaces) in the informative workspace of co-located agile (Scrum) software development teams. The study is based on the premise that displaying relevant information in a teams informative workspace is beneficial, as it readily provides information of common interest. However, existing research shows a lack of experience with software visualization tools in commercial settings. To our current knowledge, there is no study which examines ambient displays in Scrum teams with a longitudinal approach in order to address challenges in awareness and informal communication. A mixed-methods field study, which began in February 2014, is currently being conducted in cooperation with a medium-sized software company near Hamburg, Germany. Our current findings, which the following addresses, support the value of the Ambient Surfaces as a beneficial tool for Scrum teams.

QuP: Graceful Degradation in State Propagation for DVEs

Large Distributed Virtual Environments (DVEs) combine the interactions of many users, constituting a highly complex distributed system that encompasses of a multitude of heterogeneous computers and communication over varying networks. The size of a DVE is not only measured in the spatial reach of its virtual world, but also the number of concurrent users it can provide a sufficient quality service for. The existence of hardware or software failure of these many parts must be considered the normal state of operations for any sufficiently large DVE. We propose QuP, an approach to provide high fault tolerance when compared to many P2P approaches. QuP is also able provide the service of a large DVE service to clients connected via asymmetric or lower bandwidth networks.

Budget based dynamic state update aggregation

Distributed Virtual Environments (DVEs) can handle large numbers of participants by optimizing the routing of state update messages sent to the computer of the individual participant. Often Area of Interest (AoI) management is used to limit the information an avatar can observe and thus the necessary state update messages that are sent to the client which the avatar represents. If the avatars are roughly evenly distributed through out the virtual environment, AoI enables the support for very large numbers of participants. But once a sufficient number of avatars gather in a small area of the virtual space, communication overhead grows on a n2 scale as they observe each other. Instead on optimizing the overall number of messages, we limit the number of messages sent to a single client. We propose a novel approach to dynamically aggregate state update messages. It can support a very high numbers of avatars in restricted spaces. We decouple the management of the global state of the DVE from amount of detail sent to the client. Our approach exploits information about the specific DVE and its current state to maximize the ratio of bandwidth reduction to state update fidelity. It is intended to provide rich user experience for clients that connect via limited network links, like mobile users or consumer grade DSL.

Robust False Positive Detection for Real-Time Multi-target Tracking

We present a real-time multi-target tracking system that effectively deals with false positive detections. In order to achieve this, we build a novel motion model that treats false positives on background objects and false positives on foreground objects such as shoulders or bags separately. In addition we train a new head detector based on the Aggregated Channel Features (ACF) detector and propose a schema that includes the identification of true positives with the data association instead of using the internal decision-making process of the detector. Through several experiments, we show that our system is superior to previous work.

Scalable context-aware development infrastructure for interactive systems in smart environments

Context-aware systems for smart environments can be very complex and demanding for developers especially in distributed computing and communication environments. We propose a new development infrastructure, that targets this challenge by improving the general systems scalability and traceability. The infrastructure has been developed for and tested in two research labs for smart environments and human computer interaction. First measurements show that the platform has high scalability and low message latency that is perfectly suitable for interactive projects and virtual reality experiments.

Avatar Density Based Client Assignment

Scaling the number of supported users for Massive Multi-User Games (MMOGs) allows more users to experience its content together. Supporting this capability needs consider the chain of all components that constitute the system between the client software of any two users.

What’s AI by the way

ZusammenfassungEin paar persönliche Reflektionen eines Lehrers über eine Vorlesung Künstliche Intelligenz