Open Access

Advances in eye-tracking control for assistive robotic arms provide intuitive interaction opportunities for people with physical disabilities. Shared control has gained interest in recent years by improving user satisfaction through partial automation of robot control. We present an eye-tracking-guided shared control design based on insights from state-of-the-art literature. A Wizard of Oz setup was used in which automation was simulated by an experimenter to evaluate the concept without requiring full implementation. This approach allowed for rapid exploration of user needs and expectations to inform future iterations. Two studies were conducted to assess user experience, identify design challenges, and find improvements to ensure usability and accessibility. The first study involved people with disabilities by providing a survey, and the second study used the Wizard of Oz design in person to gain technical insights, leading to a comprehensive picture of findings.

Command & Control (C2) frameworks are a popular tool for bad actors to attack and infiltrate infrastructures and systems. They allow long-lasting inroads to be made into the infrastructure, through which attackers can interact with it through covert channels. These frameworks thus also play a crucial role in cybersecurity, enabling red teams and penetration testers to simulate those real-world adversary tactics. Cobalt Strike, a widely used proprietary C2 framework, offers an extensible plugin system through Beacon Object Files (BOFs). Mythic, an open-source alternative, provides a modular architecture but lacks native BOF compatibility. This thesis explores the feasibility of integrating Cobalt Strike’s BOF capabilities into a Mythic-based beacon developed at cirosec. The research begins by analyzing the structural and functional differences between Cobalt Strike and Mythic, focusing on their plugin systems and execution environments. It then examines the technical details of BOF execution, including Dynamic Function Resolution (DFR), memory management, and interactions with the beacon Application Programming Interface (API). The core contributions of this work are the design and implementation of a generic BOF runtime and the implementation of it within the Mythic-based beacon “ciroStrike” developed by cirosec. By adapting BOF execution mechanisms and ensuring compatibility with Mythic’s architecture, this integration enhances the beacon’s flexibility while maintaining its compact and evasive nature. Furthermore, an analysis of publicly available BOF implementations evaluates their applicability to this approach. The results demonstrate that BOFs can be successfully executed within Mythic with minimal modifications, bridging the gap between proprietary and open-source C2 frameworks. This research contributes to the evolution of offensive security tooling by expanding the interoperability of red team frameworks and improving the adaptability of C2 beacons.

Die Hochschule Offenburg nimmt jedes Jahr mit dem Team magmaOffenburg am RoboCup teil. Hierbei tritt das Offenburger Team in der 3D-Simulationsliga gegen Teams aus der ganzen Welt im Fußball mit autonomen Robotern an. Nun soll der Torwart von Team magmaOffenburg mit Hilfe von Deep Reinforcement Learning verbessert werden. Hierfür wird der Algorithmus PPO unter der Nutzung der RL-X Bibliothek verwendet. Im Laufe dieser Arbeit werden mehrere Versuche durchgeführt, bei denen Modelle für Torwartbewegungen entstehen. Eine dieser Bewegungen kann auf einem sehr schwer abzusichernden Bereich des Tors, in welchem der bisherige Torwart nur 5 Prozent der Bälle halten konnte, nun 36 Prozent der Bälle halten. Diese Arbeit erläutert des Weiteren die Integration dieses Modells für den Torwart in das Spiel von Team magmaOffenburg. Hierbei konnte beim Testen gegen das eigene, vorherige Team eine Reduktion von durchschnittlich 0,135 Gegentoren pro Spiel erreicht werden. Schlussendlich befasst sich diese Arbeit auch noch mit dem Weitertrainieren dieses Modells auf einem erweiterten Torbereich per Curriculum, um die Entscheidungslogik des Torwarts teilweise in das gelernte Modell zu integrieren. Hierdurch konnten in einer Spielserie jedoch nur minimal bessere Ergebnisse von 0,08 Gegentoren weniger als beim bisherigen Torwart erzielt werden.