arXiv subjects

Ege Yuceel

Publications and source records attributed to Ege Yuceel.

7 recordsLinked to original sources

arXiv2026

Factored Diffusion Policies:Compositionally Generalized Robot Control with a Single Score Network

Robotic tasks are typically specified by a tuple of factors, such as the object to be grasped, the obstacles to be avoided, the color of the target, and so on. Collecting expert demonstrations for every combination of factor values grows combinatorially. We present factored diffusion policies: a single shared diffusion network trained with per-factor null-token dropout, whose score decomposes additively across factors at inference. Under approximate conditional independence between factors given the action-observation pair, this composition approximates the true joint score with a bounded uniform error, reducing the training-task budget from a product of factor cardinalities to a sum. A trajectory-tube certificate chains this score-level bound through the reverse-time sampling ODE and a contracting tracking controller into a closed-loop state-trajectory tube whose radius factors into an ODE-sensitivity constant and a per-factor score-error budget. Unlike compositional-diffusion methods for control that combine separately trained networks, we use one shared network. Drone racing experiments confirm both the generalization bound and the certificate. On state-based multi-gate racing, the factored policy passes 90% of held-out gates -- matching an oracle -- while a K-network composition baseline collapses to 3%; on vision-based single-gate traversal, it transfers zero-shot to an unseen venue with +11.7pp success-rate gain and 2.4X crash-rate reduction.

cs.LG↗

arXiv2026

Minimal Information Control Invariance via Vector Quantization

Safety-critical autonomous systems must satisfy hard state constraints under tight computational and sensing budgets, yet learning-based controllers are often far more complex than safe operation requires. To formalize this gap, we study how many distinct control signals are needed to render a compact set forward invariant under sampled-data control, connecting the question to the information-theoretic notion of invariance entropy. We propose a vector-quantized autoencoder that jointly learns a state-space partition and a finite control codebook, and develop an iterative forward certification algorithm that uses Lipschitz-based reachable-set enclosures and sum-of-squares programming. On a 12-dimensional nonlinear quadrotor model, the learned controller achieves a $157\times$ reduction in codebook size over a uniform grid baseline while preserving invariance, and we empirically characterize the minimum sensing resolution compatible with safe operation.

eess.SY↗

arXiv2026

Active Localization of Unstable Systems with Coarse Information

We study localization and control for unstable systems under coarse, single-bit sensing. Motivated by understanding the fundamental limitations imposed by such minimal feedback, we identify sufficient conditions under which the initial state can be recovered despite instability and extremely sparse measurements. Building on these conditions, we develop an active localization algorithm that integrates a set-based estimator with a control strategy derived from Voronoi partitions, which provably estimates the initial state while ensuring the agent remains in informative regions. Under the derived conditions, the proposed approach guarantees exponential contraction of the initial-state uncertainty, and the result is further supported by numerical experiments. These findings can offer theoretical insight into localization in robotics, where sensing is often limited to coarse abstractions such as keyframes, segmentations, or line-based features.

cs.RO↗

arXiv2025

Omniscient Attacker in Stochastic Security Games with Interdependent Nodes

The adoption of reinforcement learning for critical infrastructure defense introduces a vulnerability where sophisticated attackers can strategically exploit the defense algorithm's learning dynamics. While prior work addresses this vulnerability in the context of repeated normal-form games, its extension to the stochastic games remains an open research gap. We close this gap by examining stochastic security games between an RL defender and an omniscient attacker, utilizing a tractable linear influence network model. To overcome the structural limitations of prior methods, we propose and apply neuro-dynamic programming. Our experimental results demonstrate that the omniscient attacker can significantly outperform a naive defender, highlighting the critical vulnerability introduced by the learning dynamics and the effectiveness of the proposed strategy.

cs.GT↗

arXiv2025

Performance-Guided Refinement for Visual Aerial Navigation using Editable Gaussian Splatting in FalconGym 2.0

Visual policy design is crucial for aerial navigation. However, state-of-the-art visual policies often overfit to a single track and their performance degrades when track geometry changes. We develop FalconGym 2.0, a photorealistic simulation framework built on Gaussian Splatting (GSplat) with an Edit API that programmatically generates diverse static and dynamic tracks in milliseconds. Leveraging FalconGym 2.0's editability, we propose a Performance-Guided Refinement (PGR) algorithm, which concentrates visual policy's training on challenging tracks while iteratively improving its performance. Across two case studies (fixed-wing UAVs and quadrotors) with distinct dynamics and environments, we show that a single visual policy trained with PGR in FalconGym 2.0 outperforms state-of-the-art baselines in generalization and robustness: it generalizes to three unseen tracks with 100% success without per-track retraining and maintains higher success rates under gate-pose perturbations. Finally, we demonstrate that the visual policy trained with PGR in FalconGym 2.0 can be zero-shot sim-to-real transferred to a quadrotor hardware, achieving a 98.6% success rate (69 / 70 gates) over 30 trials spanning two three-gate tracks and a moving-gate track.

cs.RO↗

arXiv2024

Strategizing against Q-learners: A Control-theoretical Approach

In this paper, we explore the susceptibility of the independent Q-learning algorithms (a classical and widely used multi-agent reinforcement learning method) to strategic manipulation of sophisticated opponents in normal-form games played repeatedly. We quantify how much strategically sophisticated agents can exploit naive Q-learners if they know the opponents' Q-learning algorithm. To this end, we formulate the strategic actors' interactions as a stochastic game (whose state encompasses Q-function estimates of the Q-learners) as if the Q-learning algorithms are the underlying dynamical system. We also present a quantization-based approximation scheme to tackle the continuum state space and analyze its performance for two competing strategic actors and a single strategic actor both analytically and numerically.

cs.GT↗

arXiv2023

Convergence of Heterogeneous Learning Dynamics in Zero-sum Stochastic Games

This paper presents new families of algorithms for the repeated play of two-agent (near) zero-sum games and two-agent zero-sum stochastic games. For example, the family includes fictitious play and its variants as members. Commonly, the algorithms in this family are all uncoupled, rational, and convergent even in heterogeneous cases, e.g., where the dynamics may differ in terms of learning rates, full, none or temporal access to opponent actions, and model-based vs model-free learning. The convergence of heterogeneous dynamics is of practical interest especially in competitive environments since agents may have no means or interests in following the same dynamic with the same parameters. We prove that any mixture of such asymmetries does not impact the algorithms' convergence to equilibrium (or near equilibrium if there is experimentation) in zero-sum games with repeated play and in zero-sum (irreducible) stochastic games with sufficiently small discount factors.

cs.GT↗