arXiv:2512.08139v1 Announce Type: new 
Abstract: The growing prevalence of artificial intelligence (AI) in various applications underscores the need for agents that can successfully navigate and adapt to an ever-changing, open-ended world. A key challenge is ensuring these AI agents are robust, excelling not only in familiar settings observed during training but also effectively generalising to previously unseen and varied scenarios. In this thesis, we harness methodologies from open-endedness and multi-agent learning to train and evaluate robust AI agents capable of generalising to novel environments, out-of-distribution inputs, and interactions with other co-player agents. We begin by introducing MiniHack, a sandbox framework for creating diverse environments through procedural content generation. Based on the game of NetHack, MiniHack enables the construction of new tasks for reinforcement learning (RL) agents with a focus on generalisation. We then present Maestro, a novel approach for generating adversarial curricula that progressively enhance the robustness and generality of RL agents in two-player zero-sum games. We further probe robustness in multi-agent domains, utilising quality-diversity methods to systematically identify vulnerabilities in state-of-the-art, pre-trained RL policies within the complex video game football domain, characterised by intertwined cooperative and competitive dynamics. Finally, we extend our exploration of robustness to the domain of LLMs. Here, our focus is on diagnosing and enhancing the robustness of LLMs against adversarial prompts, employing evolutionary search to generate a diverse range of effective inputs that aim to elicit undesirable outputs from an LLM. This work collectively paves the way for future advancements in AI robustness, enabling the development of agents that not only adapt to an ever-evolving world but also thrive in the face of unforeseen challenges and interactions.

تبرز أطروحة جديدة بعنوان 'وكلاء أقوياء في عوالم مفتوحة' أهمية تطوير وكلاء الذكاء الاصطناعي (AI) القادرين على التكيف مع بيئات متنوعة وغير متوقعة. تقدم الدراسة MiniHack، وهو إطار عمل قائم على لعبة NetHack، مصمم لإنشاء مهام متنوعة لوكلاء التعلم المعزز بهدف تعزيز قدراتهم على التعميم.

Una nueva tesis titulada 'Agentes Robustos en Mundos Abiertos' enfatiza la importancia de desarrollar agentes de inteligencia artificial (IA) que puedan adaptarse a entornos diversos e impredecibles. La investigación presenta MiniHack, un marco de trabajo basado en el juego NetHack, diseñado para crear tareas variadas para agentes de aprendizaje por refuerzo con el fin de mejorar sus capacidades de generalización.

Une nouvelle thèse intitulée 'Agents Robustes dans des Mondes Ouverts' souligne l'importance de développer des agents d'intelligence artificielle (IA) capables de s'adapter à des environnements divers et imprévisibles. La recherche présente MiniHack, un cadre de bac à sable basé sur le jeu NetHack, conçu pour créer des tâches variées pour les agents d'apprentissage par renforcement afin d'améliorer leurs capacités de généralisation.

A new thesis titled 'Robust Agents in Open-Ended Worlds' emphasizes the importance of developing artificial intelligence (AI) agents that can adapt to diverse and unpredictable environments. The research introduces MiniHack, a sandbox framework based on the game NetHack, designed to create varied tasks for reinforcement learning agents to enhance their generalization capabilities.

Robust Agents in Open-Ended Worlds

arXiv:2512.08467v1 Announce Type: new 
Abstract: Football player tracking is challenged by frequent occlusions, similar appearances, and rapid motion in crowded scenes. This paper presents a lightweight SAM-based tracking method combining the Segment Anything Model (SAM) with CSRT trackers and jersey color-based appearance models. We propose a team-aware tracking system that uses SAM for precise initialization and HSV histogram-based re-identification to improve occlusion recovery. Our evaluation measures three dimensions: processing speed (FPS and memory), tracking accuracy (success rate and box stability), and robustness (occlusion recovery and identity consistency). Experiments on football video sequences show that the approach achieves 7.6-7.7 FPS with stable memory usage (~1880 MB), maintaining 100 percent tracking success in light occlusions and 90 percent in crowded penalty-box scenarios with 5 or more players. Appearance-based re-identification recovers 50 percent of heavy occlusions, demonstrating the value of domain-specific cues. Analysis reveals key trade-offs: the SAM + CSRT combination provides consistent performance across crowd densities but struggles with long-term occlusions where players leave the frame, achieving only 8.66 percent re-acquisition success. These results offer practical guidelines for deploying football tracking systems under resource constraints, showing that classical tracker-based methods work well with continuous visibility but require stronger re-identification mechanisms for extended absences.

تم تطوير طريقة جديدة خفيفة لتتبع لاعبي كرة القدم، تجمع بين نموذج Segment Anything (SAM) ومتعقب CSRT ونماذج المظهر المعتمدة على لون القميص لتحسين استعادة الحجب. يحقق هذا النظام معدلات نجاح عالية في التتبع، حتى في السيناريوهات المزدحمة، مما يثبت فعاليته في التطبيقات الزمن الحقيقي.

Se ha desarrollado un nuevo método ligero de seguimiento de jugadores de fútbol que integra el Segment Anything Model (SAM) con rastreadores CSRT y modelos de apariencia basados en el color de las camisetas para mejorar la recuperación de oclusiones. Este sistema logra altas tasas de éxito en el seguimiento, incluso en escenarios abarrotados, demostrando su efectividad en aplicaciones en tiempo real.

Une nouvelle méthode légère de suivi des joueurs de football a été développée, intégrant le Segment Anything Model (SAM) avec des trackers CSRT et des modèles d'apparence basés sur la couleur des maillots pour améliorer la récupération des occlusions. Ce système atteint des taux de succès de suivi élevés, même dans des scénarios encombrés, démontrant son efficacité dans des applications en temps réel.

A new lightweight football player tracking method has been developed, integrating the Segment Anything Model (SAM) with CSRT trackers and jersey color-based appearance models to enhance occlusion recovery. This system achieves high tracking success rates, even in crowded scenarios, demonstrating its effectiveness in real-time applications.

Robust Agents in Open-Ended Worlds

Was this article worth reading? Share it

Chattermate

LucidQuery AI

AIvilization