Reinforcement_Learning
-
Mitigating Reward Hacking in Preference Reward Models with Evolving Penalties
2026-10-20 22:14:00Afonso Baldo, Ant\'onio Farinhas, Marcos Treviso, Ricardo Rei, Nuno M. GuerreiroAbstract
Reinforcement learning for language models has moved from preference reward models to verifiable rewards, which sustain much longer optimization. Preference models are prone to reward hacking: the policy raises their score without improving response quality, or even while degrading it. Yet for open-ended tasks, where quality can neither be verified nor fully written down as rubrics, they remain valuable. We introduce Dynalize, which mitigates reward hacking online, during policy training. Periodically and asynchronously, a language model analyzes training rollouts and writes penalties for undesirable behaviors that the reward model appears to reward; a judge applies them to every rollout. In clinical counseling and creative writing, training on the reward model alone raises its score continuously but ultimately leads to a decline in external benchmarks; Dynalize sustains longer training and results in improved downstream performance. We find that our method suppresses patterns incentivized by preference models that, without penalties, spread to most responses, and even to the model's reasoning, which the reward model does not explicitly score. Furthermore, we show that its gains hold in multi-turn training, and even after we retrain the reward model on the discovered behaviors.
Abstract (translated)
URL
https://arxiv.org/abs/2610.26496
PDF
-
SLoSa: Small, Low-Power, and Safe Foundation Models for Robotic Task Planning
2026-10-20 22:09:44Weizhe Xu, Jialiang Fan, Essien Taylor, Zijun Wang, Russell Joseph, Fanxin KongAbstract
Safe robotic task planning requires planners that can reliably achieve task goals while satisfying safety constraints. Classical planners search over explicit symbolic task representations and can provide formal guarantees with respect to encoded safety constraints, but require rigid input formats. Their computational cost also increases rapidly with the search space. In contrast, large language models (LLMs) offer flexible task interfaces and strong generalization capabilities, but they are large, power-hungry, and pre-trained with little robot-specific safety data and knowledge. Thus, they can be difficult to deploy locally on robotic systems with limited resources. To address these limitations, we present SLoSa, a neuro-symbolic compression and fine-tuning framework for training foundation-model-based robotic task planners with Small size, Low power demand, and Safety awareness. The framework first constructs a safety-aware dataset using classical planners, an advanced teacher LLM, and formal verifiers, followed by two training stages: supervised fine-tuning and reinforcement learning (RL) fine-tuning guided by a verification-based reward machine to enforce safety alignment. We evaluate SLoSa across four Planning Domain Definition Language (PDDL) domains, two household simulation environments, local edge platforms, and a desktop robotic arm. Across the PDDL tasks, verifier-guided post-training raises the success rate of compact 4B-14B planners to 70.5-76.0%, while INT4 deployment on a Jetson Orin requires 17.85-26.41 W of average power. Results in household simulators and on the physical robot further demonstrate the applicability of SLoSa across diverse robotic task settings.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28955
PDF
-
One Policy, Many Styles: Where to Aggregate Preferences for Steerable Humanoid Control?
2026-10-20 20:59:05Ziqian Huang, Yuzhuo Chen, Shimin Zhang, Junyang He, Xinyue Annie Yang, Animesh GargAbstract
Modern robotic agents are increasingly expected to support steerable control, where a single policy can adjust its behavioral style at runtime under changing user preferences. Preference-conditioned reinforcement learning provides this interface, but conditioning alone does not ensure that a requested behavior change is strong, does not leak into unrelated behaviors, and does not compromise task performance. Our evaluation combines Preference Reward Responsiveness (PRR) with coverage and utility metrics. We use these measures to compare reward-level, advantage-level, loss-level, and PCGrad-based aggregation across two humanoid-control frameworks. No aggregation level performs best across all environments and criteria. Experiments show that aggregating objectives early produces a wider range of behavioral responses but greater unintended changes in other objectives, while aggregating them later improves control precision at some cost to responsiveness.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28925
PDF
-
DualHeads: Reward-tilted few-step diffusion made simple
2026-10-20 19:31:03Yuanzhi Zhu, Eleftherios Tsonis, Simo Alami Chehboune, Xi Wang, St\'ephane Lathuili\`ere, Vicky KalogeitonAbstract
Distilled diffusion models generate high-quality images in a few sampling steps, but aligning them with task requirements and human preferences is difficult. Methods developed for multi-step diffusion models rely on differentiable rewards, or on score fields and stochastic trajectories that distilled generators may lack. We ask whether a few-step generator can be adapted using only external reward evaluations and reference samples, and show that two binary classifiers suffice. Our method, DualHeads, implements them as lightweight heads on a frozen diffusion backbone. The reward head learns which samples of the same prompt score higher, the reference head separates reference images from generated ones, and the generator maximizes a weighted sum of their logits. When the heads are optimal, this objective is KL-regularized maximization of the learned reward, and the ratio of head weights sets the regularization strength. Empirically, with both rule-based and model-based rewards, a few-step DualHeads generator reaches task rewards close to those of 40-step reinforcement learning methods while preserving image quality, at no extra inference cost. We hope that this simple and elegant dual-head formulation will help advance few-step reward-tilted generation.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28858
PDF
-
HorizonComposer: Spatiotemporally Consistent Driving Video Editing with Enriched Traffic Semantics
2026-10-20 18:51:48Mauricio Soroco, Yuqiu Liu, Zaid Tasneem, Francesco Pittaluga, Abhishek Aich, Wuyang Chen, Manmohan Chandraker, Ziyu JiangAbstract
Instruction-guided video generative models offer a scalable solution for stress-testing autonomous driving systems by simulating diverse environmental conditions. However, driving scenes operate within a dense Operational Design Domain (ODD) governed by strict traffic semantics. When applied to global driving-condition edits, general-purpose instruction-guided video editors frequently corrupt these semantics, leading to structural hallucinations and ``temporal shock.'' This work distills and protects traffic semantics during video editing at three levels. First, we construct a hybrid dataset by combining temporally aligned real-world videos with synthetic edits, balancing foundational scene structure with unambiguous appearance signals. Second, during training, we adapt advantage conditioning from offline reinforcement learning to align the model using dense rewards. We extract dense, pixel-level rewards from the hybrid data to serve as a localized signal and actively steer generation toward structurally reliable pixels. Finally, our model mitigates temporal shock by introducing thinking frames, a transition mechanism that seamlessly reconciles static image conditioning with dynamic video context. Extensive experiments demonstrate that our method prevents the corruption of multi-actor dynamics, establishing state-of-the-art spatiotemporal consistency, yielding consistent user preference (39% relative improvement in content preservation and 12% in temporal consistency), and achieving a significant 5% gain in downstream road line segmentation and 7% gain in downstream multi-object detection.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28829
PDF
-
LIFT: Scoped Constraints for Evaluating and Training Instruction Following
2026-10-20 18:40:22Eran Goldstein, Yuval Belfer, Yonatan Belinkov, Daniel GissinAbstract
Common instruction following (IF) benchmarks, such as IFEval, evaluate the ability of language models (LMs) to follow simple constraints that can easily be verified programmatically. These same constraints, in turn, are used as reward signals for reinforcement learning. This design may lead to reward hacking of constraints, pushing models away from the main task. Recent attempts to address this resulted in benchmarks with richer constraints, but required expert labor, remained closed-access, and cannot be easily extended. We present Lift-Bench, an IF benchmark that asks a model to draft a document given a topic and a structured outline. Constraints are scoped to certain document sections, providing finer control over constraint density. Lift-Bench includes both verifiable and judged constraints, as well as a training counterpart, Lift-Train. Importantly, Lift-Bench provides both main task and instruction-following constraints, mitigating shallow reward hacking. On experiments with multiple models, we find that frontier models struggle with the harder subsets of our benchmark and analyze model performance along multiple axes. We also find that training on Lift-Train is much less susceptible to reward hacking than training on the common IFTrain. Lift-Bench and Lift-Train are publicly available, as well as the data generating pipeline.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28819
PDF
-
Is Anyone Enforcing the Cartel? What Reinforcement Learning Teaches LLM Agents in Markets
2026-10-20 18:18:10Hieu Tran, Hong YuAbstract
Evaluations of large language model (LLM) agents report that pricing agents collude at others' expense. Yet collusion is defined by enforcement: a cartel, in which competitors share monopoly profits, holds only if a member who cheats is punished. We introduce an audit that tests enforcement by forcing a member to cheat in counterfactual copies of a market, and use it to study what reinforcement learning (RL) teaches agents in pricing and trading. Small LLMs trained by RL learn what their training rivals reward, but only what they readily try. In self-play, they learn fixed rules ignoring their rivals; against a punishing rival, to be deterred; against a cheating rival, to match its cuts, never the costly price war that would deter it, which they rarely try. A war forced into training is learned, but mostly only where it was forced. Against their own copies, deterred and matching agents can earn the same cartel profits, but only matching ones can enforce them, and then thinly: as a by-product and only against large cuts. Q-learners enforce collusion only in pricing. Prompted LLMs reach cartel-like outcomes mostly through rules that ignore the market. Outcome measures cannot tell these cases apart; the audit can.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28787
PDF
-
Measure First, Then Allocate: Calibrated Rollout Budgets for MCP Agents in Reinforcement Learning
2026-10-20 18:15:52Hieu Tran, Zonghai Yao, Sunjae Kwon, Hong YuAbstract
Training tool-using language models requires costly interaction, yet identical rewards within a rollout group provide no group-relative learning signal. We use reward disagreement to guide harness evaluation and the allocation of rollouts for policy training in Model Context Protocol (MCP) environments. We first introduce Discord, which tests harness edits using exact paired inference and decision-preserving early stopping: only disagreeing paired outcomes favor one harness over another. This insight motivates prospective allocation in BRACE (Bank-calibrated Rollout Allocation under Correlated Episodes). BRACE combines historical task evidence, uncertainty in task success rates, and a correlated reward model to forecast mixed-reward groups before allocating generation budget. Controlled harness experiments favor Discord, while live MCP experiments expose a retention-cost trade-off. At three Qwen3-VL model scales, BRACE exceeds the strongest plotted baseline peaks with fewer new training rollouts; historical-bank collection is counted separately. It also leads the compared methods on BFCL and NESTFUL without further training. Ablations probe the bank, estimator, and allocation score; prospective evaluation checks forecast calibration. These results support forecasting reward contrast before generation.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28785
PDF
-
Complexity-Aware Surgical Video Reasoning Segmentation with Digital Twin and Context Memory
2026-10-20 18:10:55Yiqing Shen, Mathias UnberathAbstract
Conventional surgical segmentation relies on closed-set categories or explicit text queries, which cannot resolve implicit queries that require reasoning about workflows, instrument functions, and temporal relationships. To address these limitations, we propose SurgReasonSeg, a complexity-aware method that brings reasoning segmentation over digital twin (DT) representations to surgical video. Specifically, SurgReasonSeg constructs DT representations from surgical foundation models and trains two low-rank adapters (LoRA) of one large language model with reinforcement learning (RL), each specialized for a different query complexity. A lightweight router, also trained with RL, selects between the two adapters to balance latency against accuracy. SurgReasonSeg also sets the temporal context with an adaptive sliding window conditioned on query complexity and workflow phase, and retrieves relevant historical cases from a surgical context memory by workflow similarity and procedural relevance. We further introduce EndoVis-ReasonSeg, a surgical reasoning segmentation benchmark with 1,460 training and 961 test samples across semantic, spatial, and temporal categories. On EndoVis-ReasonSeg, SurgReasonSeg improves J&F over the strongest baseline, ReSurgSAM2, by 0.178, 0.237, and 0.297 on semantic, spatial, and temporal queries, and it also achieves the best results on the Ref-EndoVis17 and Ref-EndoVis18 referring segmentation benchmarks. A backbone-matched attribution protocol separates the contribution of the upstream segmenter from that of the reasoning modules. Our method runs at 18.8 FPS averaged over the query distribution, enabling real-time reasoning segmentation in surgical videos.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28776
PDF
-
Learning Adaptable Dexterous Skills via Hybrid Dexterous Mimicking
2026-10-20 17:59:32Yuxuan Kuang, Jiashun Wang, Yixuan Sun, Yufei Ye, Sungjae Park, Umar Iqbal, Katerina Fragkiadaki, Shubham TulsianiAbstract
Dexterous multi-finger hands hold promise to unlock diverse skills like functional grasps, in-hand reorientation, and dynamic handover that the human-designed world assumes, yet existing learning recipes force a trade-off. Goal-reaching reinforcement learning generalizes to arbitrary object poses but, lacking any human prior, converges to unnatural and unreliable finger strategies; motion tracking reproduces human-like behavior from a single clip but cannot deviate from it. We present AdaDex, a goal-conditioned RL framework that turns a single hand-object trajectory of a few seconds---from mocap data or HOI reconstruction from in-the-wild RGB(D) video---into an adaptable dexterous skill. At the core of AdaDex is Hybrid Dexterous Mimicking: a shared policy is trained on two populations of massively parallel environments: motion tracking environments trained with dense motion tracking objective, which instill the human prior from human motion; and generalization environments with noised initial states and goals and a sparse object-success reward, which force the skill to work far off the reference. From seconds of human motion per skill, the resulting policies can learn diverse adaptable dexterous skills, such as functional grasping, object reposing, in-hand reorientation with finger gaiting, and dynamic behaviors like throw-and-catch. Our method outperforms baselines in terms of generalization and functionality, and we demonstrate transfer across simulators and to the real world. Project page: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28628
PDF
-
AgenticRoboRL: Bringing Coding-Agent Post-Training to Agentic Robotics
2026-10-20 17:41:09Xinxin Liu, Bingxin Xu, Wenrui Bao, Ming Li, Wentao Bao, Chen Chen, Yuzhang ShangAbstract
Agentic robotics systems combine planners with robot harnesses to perform complex manipulation tasks. Recent work improves these systems through better harnesses and external memory, but relies on powerful closed-source models as planners and does not improve the planners' own capabilities. In contrast, coding agents have improved substantially by post-training the model itself. We therefore investigate whether post-training can also improve robot planners. However, applying it across robot harnesses requires accommodating different execution mechanisms, including VLA skill invocation and control-code generation. To this end, we introduce AgenticRoboRL, a framework for post-training robot planners across different harnesses. The training recipe combines supervised fine-tuning with multi-turn reinforcement learning. To accommodate different execution mechanisms, harness-specific adapters connect recorded planner interactions with task outcomes, enabling a common RL training pipeline while keeping execution components fixed. On LIBERO-PRO, AgenticRoboRL raises the success rate of a 9B-parameter planner from 18.1% to 71.0% without external memory, approaching the 72.1% achieved by Codex (GPT-5.5) with external memory. These results demonstrate post-training as an effective approach for improving robot planners. We will release the framework and trained planner checkpoints.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28589
PDF
-
SuperMimic: Scaling Humanoid Whole-Body Control with One Differentiable World Model
2026-10-20 16:29:30Qiayuan Liao, Yiyang Shao, Junfeng Long, Kevin Zakka, Roberto Horowitz, Pieter Abbeel, Koushil SreenathAbstract
Tracking large human-motion datasets has become the dominant route to general, expressive humanoid whole-body control, but scaling across many motions is expensive. Current pipelines use model-free reinforcement learning, either merging per-motion specialists through lossy distillation, or brute-forcing a single policy with cluster-scale compute. SuperMimic amortizes learning across all motions with one differentiable world model of their shared dynamics, fit from logged transitions of a standard rigid-body simulator. The world model and the tracking policy train together in one loop, and backpropagation-through-time over short imagined rollouts gives the policy analytic gradients. Fast imagined rollouts and analytic gradients make policy learning efficient. In 3 hours on a single GPU, SuperMimic surpasses the return of a state-of-the-art motion tracker baseline trained for 7 days on 128 GPUs. It reduces GPU-hours by more than $1000\times$ and fails on only 25 of 129,514 benchmark motions, versus 179 failures for the baseline. After 9.8 hours on two GPUs, SuperMimic achieves zero failures. From 1 to 264 hours of motion, the world model's prediction error stays stable, and GPU-hours to a target return grow slowly. Our policy transfers to a physical Unitree G1 across diverse and acrobatic motions. SuperMimic removes compute as the bottleneck to scaling humanoid whole-body control, making it practical to measure motion-tracking scaling laws.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28480
PDF
-
Answering in the User's Variety: Calibrated Accommodation Pairs for Post-Training
2026-10-20 16:26:03Zhizhuo Kou, Heming Li, Jian Yang, Junyu Luo, Jingshu Peng, Sirui Han, Yike GuoAbstract
People adapt how they talk to the person in front of them. However, chat models tend to respond in Standard English regardless of the dialect used by the user. A uniform answer can hide whether training treats these users alike, and output audits cannot tell as well. We trace how each post-training stage, from supervised fine-tuning (SFT) to preference optimization (DPO) and reinforcement learning, changes a model's tendency to match the user, and read what each update prefers from two content-matched replies, one matching the user and one not. Across 10 open model lineages, post-training standardizes replies to every informal cue, yet after adjusting for how likely each reply already was, the DPO update still favours matching British and Australian spelling over AAE and Singlish on 10 lineages, and models still produce AAE features when asked. We then introduce calibrated accommodation pairs (CAP), preference data that reward matching users' dialect density while preserving Standard English for Standard-English users. On a fully open model, CAP raises convergence on features it was not trained on without a detectable increase in dialect for Standard-English users or a cost in instruction following. In contrast, one-sided pairs reach a similar gain only by overshooting the user and imposing dialect. Our findings show that uniform replies can hide selective updates and audits should read the update as well as the output. Code: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28474
PDF
-
Fast Motion Retargeting Pipeline for Scalable Humanoid Robot Learning
2026-10-20 16:22:00Daniel Marew, Sangjun Park, Nisal Perera, Donghyun KimAbstract
While large-scale human motion capture datasets provide a rich source of demonstrations for humanoid robot learning, retargeting the diverse motions to different humanoid robots remains a nontrivial challenge. Existing kinodynamic trajectory optimization pipelines are computationally expensive and often require manual adjustments per motion. We present FRDV (Fast Retargeting with Dynamics Validation), a scalable framework consisting of rapid constrained kinematic retargeting and an additional dynamics validation pipeline. Its kinematic stage integrates a custom Sequential Quadratic Programming solver, Real-Time Iteration, and parallel chunk-based processing, achieving up to 9,156 frames per second, approximately up to 136 times faster than GMR and ProtoMotions, while robot-specific constraints are incorporated into the optimization to reduce common retargeting artifacts, such as ground penetration and self-collision. For motions requiring additional verification, a contact implicit iLQR controller is utilized to test the retargeted reference under full rigid body dynamics. This dynamics-based validation identifies dynamically infeasible motions and refines contact-related artifacts such as foot floating and foot skating. We evaluate retargeting quality by training motion tracking reinforcement learning policies using several state-of-the-art pipelines and comparing their tracking fidelity and success rates. Furthermore, sim-to-real transfer is demonstrated on a Unitree G1 humanoid. Videos and code are available at: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28470
PDF
-
FALCA: Falsification-Aware Reinforcement Learning with Hierarchical Credit Assignment for Long-Horizon Deep Search Agents
2026-10-20 15:24:44Zijun Dou, Kun Luo, Chaofan Li, Shuqi Lu, Zheng LiuAbstract
Long-horizon deep search requires LLM agents to iteratively decompose complex information needs, gather evidence from external sources, and maintain a reliable research state under multiple constraints. A central challenge is that early mistakes can be amplified over the search trajectory: agents may anchor on an incorrect candidate, collect partially supporting evidence, overlook contradictions, and accumulate errors that misguide subsequent reasoning. We propose FALCA, a falsification-aware framework that formulates deep search as a propose-falsify process. FALCA adopts a collaborative Searcher-Falsifier architecture, where the Searcher proposes candidate answers or sub-answers, while independent Falsifiers actively seek counterevidence, check unmet constraints, identify conflicts, and produce concise falsification reports for research-state revision. To strengthen this loop, we introduce falsification-aware training, combining SFT warmup on falsification trajectories with agentic RL based on parallel rollouts and hierarchical credit assignment over research stages, state transitions, and actions. Experiments on five search benchmarks show that FALCA improves long-horizon deep search performance over backbone and harness ablations, while reducing confirmation bias and error accumulation through more effective falsification and research-state revision.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28361
PDF
-
Reinforced Iterative Learning Control for elastic joint arms: Guaranteed cross-trajectory generalization under model uncertainties
2026-10-20 15:00:55Yuri De Santis, Michele Pierallini, Franco Angelini, Manolo GarabiniAbstract
Trajectory tracking for compliant robots is challenging due to nonlinear dynamics and modeling uncertainties, often requiring learning-based methods to avoid explicit parameter identification. Iterative Learning Control (ILC) ensures theoretically high precision and preserves the passive compliance of the robot. However, traditional ILC requires a separate learning phase for each new trajectory. Conversely, Reinforcement Learning (RL) generalizes to new trajectories, but practically suffers from high sample complexity and limited tracking precision. This work proposes Reinforced Iterative Learning Control (RILC) that combines the strengths of both approaches to achieve fast adaptation and guaranteed tracking convergence across diverse trajectories. By establishing a concurrent coupling between the two controllers, RILC overcomes the high initial errors and slow adaptation of traditional ILC, enabling high-performance tracking from the very first trials. In this architecture, ILC provides high-quality samples, while RL contributes as a generalizing best-effort policy across iterations. Additionally, a theoretical proof is presented to show that the RL component preserves ILC stability and, under specific conditions, can also reduce the error bound asymptotically. The efficacy of this framework is demonstrated through simulation and experiments on an underactuated 4-degree-of-freedom serial elastic robot using different trajectories and modeling mismatches.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28329
PDF
-
Reward Misspecification Degrades Chain-of-Thought Monitorability
2026-10-20 14:45:29Tianyi Lorena Yan, John HewittAbstract
Monitoring the chain-of-thought (CoT) of language models (LMs) is a key strategy for detecting and preventing misbehaviors for AI safety. As such, it is critical to identify reinforcement learning (RL) practices that lead to unmonitorable chain-of-thought. One such practice may be unavoidable: reward misspecification. As RL environments become more complex, it is infeasible to have perfect verification, and models are rewarded for behaviors that are different from the true task intent. We show that reward misspecification alone already degrades monitorability. We create a testbed with pairs of true reward aligned with task intent and misspecified rewards that favor a different behavior, so that reward hacking can be induced by training and labeled verifiably. Across three models and three tasks, CoT monitorability remains low as models' hack rates increase: the CoTs keep reasoning about the task intent while the outputs violate the intent and get rewarded. To mitigate unmonitorability without prior knowledge of the true reward or the hack, we collect rollouts with diverse rewards across training steps and use a LM to infer a reward description. We then inject the description into the input, which teaches the model to verbalize the reward. This reward description training improves monitorability by an average of 24% or 42% absolute, depending on the metric, across all models and tasks, and yields CoTs that reason about the misspecified reward explicitly. Code is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28303
PDF
-
Reframing Object-Centric Reinforcement Learning: From Absolute to Relative Representations
2026-10-20 13:38:23Cedric Derstroff, Jannis Bl\"uml, Janosch Moos, Michael Girstl, Zihan Ye, Debora Clever, Kristian KerstingAbstract
Object-centric reinforcement learning agents typically receive each object's position in a fixed global frame, even though their decisions usually depend on where objects are relative to one another. These absolute views often fail to capture the local context that is most relevant to the agent's behavior. As a result, an agent must relearn the same response at every location where a configuration can occur, and its performance collapses when the scene is merely shifted. In this work, we propose a shift toward Object-Relative Coordinates fixed to Anchor (ORCA), where entities are described in a relative frame of reference. To operationalize this idea, we introduce Object-Relative inductive-Bias Transformers (ORBiT), a transformer-based architecture for processing anchor-relative inputs in reinforcement learning. Our results suggest these relative representations improve performance for both MLP and transformer agents. Furthermore, we demonstrate that ORBiT outperforms the pixel-based PPO baseline, achieving an aggregate HNS IQM greater than 1.0 over 13 Atari games. ORBiT learns the game of Foosball from scratch, dominating all baselines. In our controlled Tag environment, we further show that ORCA extends to rotations, that ORBiT generalizes to unseen numbers of distractors, and that these gains require a decision-relevant anchor.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28197
PDF
-
Interruption-Aware Recurrent Reinforcement Learning for UAS Navigation Under Temporary Overrides
2026-10-20 12:30:41Christopher Conrad, Deepak Panda, Weisi Guo, Antonios TsourdosAbstract
Safe autonomous navigation of UASs requires tactical deconfliction to address conflicts that strategic coordination cannot prevent. This task is challenging under partial observability, when sensor or traffic data are missing, delayed, or degraded. Geometry-based methods can offer separation assurances but face limitations in dense encounters. Recurrent reinforcement learning has therefore been proposed to learn navigation policies that integrate observation-action histories to optimise long-term safety-efficiency trade-offs. However, operational safeguards may require a remote pilot, onboard detect-and-avoid system, or runtime assurance mechanism to temporarily override the learned policy. A research gap remains in understanding how these interruptions affect recurrent policies when the agent regains authority. Benchmark evidence shows that interruptions make task variables less decodable from the recurrent hidden state and increase post-interruption collision risk. The findings suggest that altered recurrent context prolongs post-handback degradation through an action-observation feedback loop, in which changed actions alter the flight path and subsequent recurrent updates. To address this vulnerability, we propose three interruption-aware architectures that condition recurrent processing on interruption context through separate feature streams, dedicated interruption memory, or gated environmental features. Across simulations, all three architectures preserve more decodable task information than a standard recurrent policy. Moreover, the strongest architecture reduces post-interruption collisions by 74% and interruption-induced mission failures by 69% relative to the recurrent baseline, without degrading interruption-free performance. These results show that safe interruption does not guarantee safe handback, and support interruption-aware recurrent design for resilient UAS autonomy.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28121
PDF
-
TrajEvolve: Trajectory Evolution for Reinforcement Learning with Hindsight Credit Assignment
2026-10-20 11:48:01Yi Wen, Hao Chen, Wanyu Wang, Maolin Wang, Pengyue Jia, Derong Xu, Huize Tan, Yingyi Zhang, Wenlin Zhang, Weihong Luo, Xiku Du, Xiangyu ZhaoAbstract
The Reinforcement Learning (RL) paradigm has achieved remarkable success in enhancing the reasoning capabilities of Large Language Models (LLMs). However, when applied to long-horizon tasks, it faces severe challenges: i) Due to the high task complexity, the model struggles to sample enough successful trajectories, resulting in slow convergence. ii) Relying solely on sparse final rewards causes all steps to be updated indiscriminately, encouraging redundant step generation. When both the quantity and quality of successful trajectories become difficult to guarantee, the performance is severely constrained. This dilemma stems from the limitations: existing RL paradigms rely entirely on the reasoning capability of base models, lacking the ability to actively optimize trajectories. To address the above issues, we propose a novel trajectory evolution paradigm with hindsight credit assignment, termed TrajEvolve. TrajEvolve not only relies on the trajectory inferred by the model but also generates ``evolutionary trajectories'' by refining low-quality steps in historical trajectories, thereby exposing the policy to high-quality compositions that lie within the base model's support but are rarely sampled within practical training budgets. Furthermore, TrajEvolve reformulates step-level continuous credit assignment as a binary classification task, whether each step is necessary for task completion. This simplification not only reduces the learning difficulty but also enables step labels to be obtained through counterfactual verification, achieving precise identification of redundant steps. Extensive experiments on two datasets demonstrate the superiority of the proposed method.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28060
PDF
