Work and projects

Hochschule Bonn-Rhein-Sieg

KEROL (2024-)

Since September 2024, I lead this project that is connected to an H-BRS research grant. Here, I focus on two aspects: (i) data collection and annotation for robot skill execution, where the particular objective is to collect both nominal data and data containing various execution failures, and (ii) simulation-based testing of learned robot skills in varied scenarios.

Lecturer (2023-)

Since October 2023, I have a lecturer position in the Institute for AI and Autonomous Systems at H-BRS. Here, I teach various courses in the Master of Autonomous Systems program, in particular two mandatory courses (Autonomous Mobile Robots and Software Engineering for Robotics), as well as two elective courses (Robot Learning and Cognitive Robotics). My course materials can be found here.

In addition to my teaching duties, I still maintain a subset of my duties in b-it (see below), in particular the co-organisation of the foundations course, the PhD seminar organisation, as well as the maintenance of some student-related workflows.

MigrAVE (2021-2023)

Between the end of 2020 and 2023, I worked as a half-time research associate in the MigrAVE project, funded by the German Federal Ministry of Education and Research. In this project, I was leading the development of a robot assistant for treatment of children who have been diagnosed with Autism Spectrum Disorder (ASD), where our target group included children who may come from a family with a migration background. In contrast to some of my other projects, this one had a stronger focus on social robotics and was considerably more interdisciplinary; particularly the interdisciplinarity has been quite useful for expanding my research focus to aspects dealing with human-robot collaboration, and concretely robot personalisation.

The components we developed in the project are available on GitHub, and videos illustrating various components developed during the work can be found on our YouTube channel.

b-it (2020-2023)

Between the beginning of 2020 and October 2023, I worked as a half-time research associate in the Bonn-Aachen International Center for Information Technology (b-it). In particular, I was an assistant to the b-it director at Hochschule Bonn-Rhein-Sieg, such that my colleagues and I managed and organised the Master of Autonomous Systems program. My position was a good mixture of administrative duties and research activities primarily related to my PhD project. Some of my activities in this position included:

• Teaching support in various courses in the program
• Regular activities related to the the b-it-bots@Home team (see below)
• Organisation of a regular PhD seminar for the PhD students in the Autonomous Systems Group (that, of course, included myself)
• Co-organisation of the foundations course for new students of the master’s program
• Maintenance of various student-related workflows, particularly those performed over LEA, the university’s course management system

b-it-bots@Home (2017-2022)

Since roughly the summer of 2017 roughly until I completed my PhD, I was part of the b-it-bots@Home team at Hochschule Bonn-Rhein-Sieg (I was team co-leader until February 2019, but had to step down for a few months due to time constraints, although I continued again in September 2019). I was focusing on various aspects in the team, but my primary activities were knowledge engineering and system integration. There was an overlap of what I was doing in the team and my PhD.

At the beginning of 2018, we acquired a Toyota HSR robot, such that I oversee most of the activities with this robot. For a while, I was trying to work with our old Care-O-bot 3 as well, but unfortunately that one got broken.

I have developed and contributed to various open-source projects here, particularly:

• mas_domestic_robotics - our primary robot-independent domestic robotics repository
• mas_execution_manager - a library for (i) specifying reusable state machine-based scenarios and (ii) defining various common interfaces for implementing robot behaviours
• mas_knowledge_base - a library that (i) exposes an rdflib-based interface for interacting with OWL ontologies and (ii) provides a wrapper for the ROSPlan knowledge base as well as mongodb_store
• action-execution - an under-development package that allows creating simple models of action execution-specific aspects
• a collection of tutorials for using the functionalities we develop in the team

ROPOD (2017-2020)

Between 2017 and April 2020, I was a half-time research associate in the ROPOD EU-funded project (the final review of the project was on April 23, 2020). My primary work was in a work package focusing on fault detection and fault tolerance; in this context, my colleagues and I were working on data logging (in particular, a concept for a so-called robotic black box), component and system monitoring, as well as remote monitoring, experimentation, and diagnosis. I also got to do quite a bit of testing with the ROPOD platform. Additionally, I was involved in task planning and a fair amount of system integration.

Most of the components that we have developed in the project are open-source; of these, I was personally mostly involved in the development of:

• our black box components (the data logging component as well as a library for working with black box data)
• a component monitoring library
• a library for defining and managing robot experiments
• a web application for remote robot operation monitoring and diagnosis
• a library for implementing fault-tolerant state machines
• components for elevator navigation (a component for managing the workflow of taking an elevator and a component for detecting floor changes using pressure sensor measurements)
• generic common interfaces for communication and component implementation
• a Gazebo-based OpenAI gym environment for the ropod
• a Python-based task planner and knowledge base interface

During the project, I also developed various minor tools that have been useful in different contexts:

• utilities for calibrating smart wheels
• a simple component that aligns a robot with a corner as soon as it starts drifting away (this has been useful for letting our robot perform simple motions, such as moving in a circle - for prolonged periods of time without worrying that it will drift away from a designated area)

beyondSPAI (2016)

After finishing my master’s thesis in 2016, I was briefly involved in the beyondSPAI project. My main activity here was performing a comparative analysis of (mostly learning-based) person detection and segmentation models and identifying candidate models that can use skin masks as additional source of information that aids the detection process.

AICISS (2014-2015)

During part of my master’s studies, I worked on the AICISS project, where I mostly worked on fault detection (which was also my R&D project during my studies). We worked with a KUKA youBot in this project, such that I looked into wheel fault detection, using a camera observation system for keeping the robot inside a predefined safe region, using fault injection for diagnosing laser scanner and wheel faults, as well as implementing a custom-made data logger.

This was the first larger robotics project I was involved in, so I gained a lot of invaluable experience in it.