Conversational Interfaces for Explainable AI: A Human-Centered Approach

in Proceedings of EXTRAAMAS 2019 (2019)

One major goal of Explainable Artificial Intelligence (XAI), in order to enhance trust in technology, is to enable the user to enquire information and explanation about its functionality directly from an ... [more ▼]

One major goal of Explainable Artificial Intelligence (XAI), in order to enhance trust in technology, is to enable the user to enquire information and explanation about its functionality directly from an intelligent agent. We propose conversational interfaces (CI) to be the perfect setting, since they are intuitive for humans and computationally processible. While there are many approaches addressing technical issues of this human-agent communication problem, the user perspective appears to be widely neglected. With the purpose of better requirement understanding and identification of implicit expectations from a human-centered view, a Wizard of Oz experiment was conducted, where participants tried to elicit basic information from a pretended artificial agent (What are your capabilities?). The hypothesis that users pursue fundamentally different strategies could be verified with the help of Conversation Analysis. Results illustrate the vast variety in human communication and disclose both requirements of users and obstacles in the implementation of protocols for interacting agents. Finally, we infer essential indications for the implementation of such a CI. [less ▲]

Model-Driven Interaction Design for Social Robots

in 4th International Workshop on Model-driven Robot Software Engineering, Marburg, Germany, 2017 (2017, December)

Robotic software development frameworks lack a possibility to present,validate and generate qualitative complex human robot interactions and robot de-velopers are mostly left with unclear informal project ... [more ▼]

Robotic software development frameworks lack a possibility to present,validate and generate qualitative complex human robot interactions and robot de-velopers are mostly left with unclear informal project specifications. The devel-opment of a human-robot interaction is a complex task and involves different ex-perts, for example, the need for human-robot interaction (HRI) specialists, whoknow about the psychological impact of the robot’s movements during the in-teraction in order to design the best possible user experience. In this paper, wepresent a new project that aims to provide exactly this. Focusing on the interac-tion flow and movements of a robot for human-robot interactions we aim to pro-vide a set of modelling languages for human-robot interaction which serves as acommon, more formal, discussion point between the different stakeholders. Thisis a new project and the main topics of this publication are the scenario descrip-tion, the analysis of the different stakeholders, our experience as robot applicationdevelopers for our partner, as well as the future work we plan to achieve. [less ▲]

Model-Based Specification, Deployment and Adaptation of Robot Perception Systems

Doctoral thesis (2017)

As robots are becoming ubiquitous and more capable, the need for introducing solid robot software development methods is pressing to increase robots' task spectrum. This thesis is concerned with improving ... [more ▼]

As robots are becoming ubiquitous and more capable, the need for introducing solid robot software development methods is pressing to increase robots' task spectrum. This thesis is concerned with improving software engineering of robot perception systems. The presented research employs a model-based approach to provide the means to represent knowledge about robotics software. The thesis is divided into three parts, namely research on the specification, deployment and adaptation of robot perception systems. The first part contributes the design and development of two domain-specific languages, namely RPSL and DepSL. Those languages provide suitable notations and abstractions to enable domain experts to express, compose and explore functional, architectural and deployment design decisions of robot perception systems. The resulting models are interpretable, thus they can be used not only to communicate design decisions to stakeholders, but also to verify them in an early development stage. The second part contributes means for deploying perception systems on real robot systems even in the presence of varying resource conditions. To this end, functional, architectural and deployment models are composed in a graph-structure. Such a graph enables not only humans, but also robots to derive implicitly defined information about their software both at design time and run time. The second part also contributes a reference architecture for deploying robot perception systems. The architecture provides a template solution for integrating not only the models required for deployment, but also all the other means required to carry out deployment. The third part utilizes both RPSL, DepSL and the reference architecture to specify, implement and evaluate three different robot perception systems. Those are capable to satisfy changing requirements induced, for example, by the robot's tasks or environment. This is achieved by proposing algorithms which derive adaptation actions based on models and varying requirements. [less ▲]

Exploring Gridmap-based Interfaces for the Remote Control of UAVs under Bandwidth Limitations

in Companion of the 2017 ACM/IEEE International Conference on Human-Robot Interaction (2017)

Graph-based Software Knowledge: Storage and Semantic Querying of Domain Models for Run-Time Adaptation

in IEEE International Conference on Simulation, Modeling, and Programming for Autonomous Robots SIMPAR, San Francisco, Dec 2016 (2016, December)

Software development for robots is a knowledge intensive exercise. To capture this knowledge explicitly and formally in the form of various domain models, roboticists have recently employed model-driven ... [more ▼]

Software development for robots is a knowledge intensive exercise. To capture this knowledge explicitly and formally in the form of various domain models, roboticists have recently employed model-driven engineering (MDE) approaches. However, these models are merely seen as a way to support humans during the robot's software design process. We argue that the robots themselves should be first-class consumers of this knowledge to autonomously adapt their software to the various and changing run-time requirements induced, for instance, by the robot's tasks or environment. Motivated by knowledge-enabled approaches, we address this problem by employing a graph-based knowledge representation that allows us not only to persistently store domain models, but also to formulate powerful queries for the sake of run time adaptation. We have evaluated our approach in an integrated, real-world system using the neo4j graph database and we report some lessons learned. Further, we show that the graph database imposes only little overhead on the system's overall performance. [less ▲]

Context-based Selection and Execution of Robot Perception Graphs

in 20th IEEE International Conference on Emerging Technologies and Factory Automation (ETFA'15) (2015, September)

To perform a wide range of tasks service robots need to robustly extract knowledge about the world from the data perceived through the robot’s sensors even in the presence of varying context-conditions ... [more ▼]

To perform a wide range of tasks service robots need to robustly extract knowledge about the world from the data perceived through the robot’s sensors even in the presence of varying context-conditions. This makes the design and devel- opment of robot perception architectures a challenging exercise. In this paper we propose a robot perception architecture which enables to select and execute at runtime different perception graphs based on monitored context changes. To achieve this the architecture is structured as a feedback loop and contains a repository of different perception graph configurations suitable for various context conditions. [less ▲]