Constrained RL

RL: Reinforcement Learning

Follow this topic Ask about this topic

Momentum

22 papers in the last four weeks, up 175% on the four weeks before. 0.2% of all new papers.

Jul 13Week of Sep 28

Latest papers 143

Jun 30, 2026cs.CR

Certified Speculative Execution for Untrusted AI Agents

Hard-constrained sequential decision systems have no certified way to spend the test-time compute of modern AI: executing the multi-step drafts of a learned policy or a frozen LLM forfeits the feasibility guarantee a trusted solver provides, while invoking the solver at every step forfeits the speed the AI offers. Certificate-Gated Prefix Acceptance (CGPA) closes this gap with a certified speculative-execution contract for untrusted AI agents: a trusted verifier rejects constraint-violating transitions exactly, a conformally calibrated value boundary gates the longest low-cost prefix within a per-segment regret budget, and the rest defers to the solver, so safety, regret, and speed decouple by construction. The contract drives every untrusted proposal source - adversarial drafters and six heterogeneous frozen LLMs (including a 12B model that violates constraints in 98% of direct rollouts) - to zero applied violations; a certificate-aware learned boundary, conformally calibrated, drives mean regret three orders of magnitude below unguarded acceptance, to within sampling noise of the stepwise oracle (95% CI spanning zero), and under calendar shift a learned proposal source overtakes it on 15 of 18 held-out days. On a deployment-scale unit-commitment instance it turns a frozen 8B LLM into a 2.96x per-episode wall-clock speedup at 2.1% regret, outpacing the domain heuristic (1.79x) and a safe receding-horizon baseline (1.07x): the more capable the untrusted source, the faster the certified system, at guarantees that never change.

Chenyu Zhou, Qiliang Jiang, Shuning Wu +1

School of Engineering, Institute of Science Tokyo, Japan · College of Control Science and Engineering, Zhejiang University, China · Department of Electrical and Computer Engineering, National University of Singapore, Singapore

AI Agent Reliability Constrained RL AI Agent Safety Runtime Enforcement for AI Agents

Jun 26, 2026cs.RO

PPO-EAL: Exact Augmented Lagrangian Proximal Policy Optimization for Safe Robotic Control

Reinforcement learning (RL) has emerged as a promising solution to accomplish complex robotic control tasks; however, most of the current work ignores the safety requirements. Safe RL seeks to maximize task performance while satisfying explicit physical constraints, but current algorithms struggle to learn the policy efficiently with precise constraint satisfaction. This work proposes PPO-EAL, a novel first-order constrained policy optimization framework that integrates exact augmented Lagrangian optimization into proximal policy optimization for safe robotic control. By combining clipped policy updates with exact quadratic penalty terms, PPO-EAL achieves theoretically grounded constraint enforcement without requiring impractically large penalty factors. A momentum-regulated multiplier update further improves dual-variable stability, reducing constraint oscillation and unsafe behavior while preserving task performance. We provide exactness and convergence analysis under standard stochastic approximation assumptions. Extensive validation across diverse GPU-accelerated robotic benchmarks-including cart-pole balancing, cart-double-pendulum stabilization, 7-DoF Franka end-effector reaching, and quadrupedal locomotion-demonstrates superior safety precision and reward performance compared with state-of-the-art first-order safe RL baselines. Finally, we demonstrate zero-shot sim-to-real deployment in a contact-rich gear assembly task, where PPO-EAL substantially improves task success, reduces peak contact force, and enhances operational robustness. These results establish PPO-EAL as a general and practically deployable safe RL framework for diverse safety-critical robotic systems.

Jiatao Ding, Songqun Gao, Andrea Del Prete +1

Department of Industrial Engineering, University of Trento, Via Sommarive 9, 38123, Trento, Italy

Robotic Control Constrained RL Robot Safety Proximal Policy Optimization Safe RL

Jun 25, 2026cs.RO

Support-Constrained RL Enables Real-World Policy Improvement without Real-World Experience

Robots trained on real world data tend to be imprecise, slow, and brittle to perturbations. Improving these policies with reinforcement learning (RL) is an appealing alternative, but this process often requires expensive training in the real world. Performing policy improvement in simulation instead provides a far cheaper alternative, but unconstrained RL in simulation can exploit contact and dynamics mismatches, resulting in unsafe behaviors that do not transfer to hardware. Common forms of regularization can furthermore limit improvement by overconstraining to an imperfect behavior prior. In this work, we propose Support-Constrained Off-Domain REinforcement (SCORE), a real-to-sim-to-real framework that constrains RL in simulation to the support of a generative policy pretrained on real data. We instantiate this constraint through flow steering, restricting SCORE to actions the base policy can already produce, which ensures transferable behaviors while maximizing policy improvement. Improving a policy with SCORE requires minimal effort: it learns from sparse rewards, avoids distillation, and leaves the base policy untouched. Across eight real-world dexterous multi-fingered robotic manipulation tasks, SCORE improves average success rate from 37.8% to 89.9%, compared to 59.5% for the best baseline, and reaches success in 36.8% fewer steps than the base policy. Ultimately, through extensive experiments and ablations, we show that simulation can substantially improve real-world manipulation policies when policy optimization is appropriately constrained, introducing a new paradigm for real-to-sim-to-real policy improvement. Videos and code are available at https://weirdlabuw.github.io/score/.

Raymond Yu, William Huey, Mustafa Mukadam +2

University of Washington · Amazon FAR

RL for Robotics Robot Policy Adaptation Robot Skill Learning Robotics Simulation Sim-to-Real Transfer Constrained RL+1

Jun 24, 2026cs.RO

Average-Power-Budgeted Underwater Vehicle Control via Constrained Reinforcement Learning

Underwater vehicles operate from a fixed onboard energy budget that propulsion rapidly depletes, so a controller that completes its task while drawing less thruster power directly extends mission range and endurance. Reinforcement learning yields capable model-free controllers for station-keeping and trajectory tracking, but optimizing task accuracy alone drives the policy toward oscillatory, energy-wasting actuation. The established remedy subtracts an energy penalty from the reward, yet this sets the task-power trade-off through a single weight with no physical units: a target power level cannot be specified, the weight must be re-tuned for every vehicle and task, and a mismatched weight can even raise power. This paper instead formulates energy-efficient underwater control as a constrained Markov decision process in which average thruster power is subject to an explicit budget, solved with a PPO-Lagrangian algorithm. The power level is set by declaring a budget in physical units, and a single dual variable is updated online to meet it for each vehicle and task, without manual weight search. Across three vehicles and four tasks in the MarineGym simulator, the energy-constrained policy draws the least power in all twelve settings, reducing it by 14--65% (up to 64.9%) over a task-only baseline and below an energy-reward baseline everywhere, while remaining the smoothest in ten settings and preserving task accuracy except in one deliberately power-limited regime. Imposing energy as an explicit constraint thus offers a tuning-free route to energy-efficient underwater control that needs no per-vehicle, per-task weight search.

Yinuo Wang, Gavin Tao, Yuze Liu +1

College of Graduate and Professional Studies, Trine University, Angola, IN 46703, USA · School of Computer Science and Statistics, Trinity College Dublin, Dublin, D02 PN40, Ireland · School of Engineering, Swinburne University of Technology, Melbourne, VIC 3122, Australia +1

RL Control Constrained RL Underwater Robotics

Jun 23, 2026cs.LG

Bias-Controlled Primal-Dual Natural Actor-Critic: Optimal Rates for Constrained Multi-Objective Average-Reward RL

Many reinforcement learning (RL) problems in the infinite-horizon average-reward setting require optimizing multiple conflicting objectives while satisfying multiple safety constraints. A common approach is concave scalarization, where the agent maximizes a utility

f(J^π_{r_1}, \ldots, J^π_{r_M})

subject to a scalarized constraint

g(J^π_{c_1}, \ldots, J^π_{c_N}) \ge 0

, where

J^π_{r_m}

and

J^π_{c_n}

denote the average-reward and cost under policy

π

. However, the nonlinearity of

f

and

g

introduces bias in policy-gradient and actor-critic methods, since gradients must be evaluated using noisy estimates of

J^π,

and

\mathbb{E}[\partial f(J^π)] \neq \partial f(\mathbb{E}[J^π]),

and this bias propagates through both primal and dual updates. We propose an MLMC-based primal-dual Natural Actor-Critic algorithm for average-reward MDPs that controls bias in scalarized objectives, constraint evaluation, and actor-critic estimation without requiring mixing-time knowledge. We show that the algorithm achieves optimal global convergence and constraint-violation rates of

\tilde{O}(1/\sqrt{T})

. To our knowledge, this is the first result establishing optimal convergence for concave scalarized multi-objective RL in the average-reward setting, both with and without constraints, and the first to do so without mixing-time information even in the absence of scalarization.

Ankur Naskar, Swetha Ganesh, Vaneet Aggarwal

Indian Institute of Science, Bengaluru, India · Purdue University, USA

Multi-Objective Reinforcement Learning Reinforcement Learning Average-Reward RL Constrained RL Actor-Critic Methods Policy Learning

Jun 22, 2026cs.LG

Learning to Trigger: Reinforcement Learning at the Large Hadron Collider

High-throughput scientific facilities such as the Large Hadron Collider depend on real-time event filtering (\textit{triggering}) under tight constraints on bandwidth, latency, and storage. In practice, trigger menus are largely static and hand-tuned and can become suboptimal as detector conditions, pileup, and background composition drift over time. We cast online threshold tuning as a sequential decision-making problem: a reinforcement learning agent ingests streaming summaries of recent rates and signal-sensitive features and updates trigger thresholds to maximize signal efficiency while tracking a target background rate within a tolerance band. We adapt Group-Filtered Policy Optimization (GFPO) to streaming control and introduce two variants (GFPO-F, GFPO-FR) that enforce background rate feasibility during training. On a benchmark that emulates realistic collider operation, we study two representative triggers: a total transverse energy (

H_{T}

) trigger sensitive to pileup variation, and an anomaly-detection (AD) trigger based on reconstruction loss for rare or non-standard signatures. On Monte Carlo streams, our agent increases the fraction of in-tolerance time intervals by 48% (

H_T

) and 28% (AD), with a cumulative gain of up to 2% in signal efficiency on those in-tolerance intervals. Transferring from simulation to \emph{real} collision data (CMS Run 283408), the same agent, without fine-tuning, achieves a 56% (

H_T

) and 28% (AD) in-tolerance improvement over baselines, with further signal-efficiency gain on both triggers. To our knowledge, this is the \emph{first} demonstration of RL-based trigger control on real Large Hadron Collider collision data. Code is available at https://github.com/Zixind/GFPO_LHC (see repo for details).

Zixin Ding, Shaghayegh Emami, Giovanna Salvi +7

University of Chicago · University of Michigan, Ann Arbor · Fermi National Accelerator Laboratory

RL Control Non-Stationary RL Constrained RL High-Energy Physics

Jun 20, 2026astro-ph.IM

Deep RL for Fast Long-Horizon Operations Scheduling on NASA's Carruthers Geocorona Observatory Mission

Spacecraft operations scheduling is a highly constrained, long-horizon combinatorial optimization problem that traditionally relies on heuristics, constraint programming, or manual planning. We present a scalable deep reinforcement learning framework developed and deployed for NASA's Carruthers Geocorona Observatory mission. Our framework introduces a macro-action abstraction known as activity blocks coupled with dynamic action-masking to navigate the intractably large search space and strictly enforce complex power, thermal, and instrument constraints. The resulting architecture generates globally feasible schedules with overwhelming probability, establishes operational trust, and executes a full training cycle in under six hours, circumventing the need for policy robustness by enabling rapid, on-demand retraining. Further, resulting schedules outperform baseline heuristics in scheduled science quality. The deep reinforcement learning framework was deployed as the default operational scheduler for the Carruthers Geocorona Observatory mission from the outset of the mission, demonstrating that deep reinforcement learning can be trusted for real spacecraft operations under complex, evolving constraints.

Alex Zhang, Jackson Craig, Lara Waldrop

Department of Electrical and Computer Engineering, University of Illinois Urbana-Champaign, IL, USA

Reinforcement Learning Long-Horizon Planning Constrained RL

Jun 18, 2026cs.AI

A Multi-Agent system for Multi-Objective constrained optimization

Many decision-making problems in computing and networking systems can be naturally formulated as cost-minimization problems under performance constraints. In dynamic environments, reinforcement learning (RL) is often used to solve such problems at runtime by embedding both costs and constraint violations into a single scalar reward through weighted penalty terms, following a Lagrangian-inspired formulation. However, in this context the behavior of the learned policy critically depends on the choice of these weights, which are typically selected manually. This makes it difficult to identify an appropriate trade-off between optimizing the primary objective and effectively avoiding constraint violations, particularly in non-stationary environments where their relative importance may change. This paper presents MAMO (Multi-Agent system for Multi-Objective constrained optimization), an approach to tackle this balancing problem through multi-agent RL. MAMO decouples task execution from objective design by formulating the selection of reward weights as a learning problem, providing a !rst step towards more autonomous and robust RL-based solutions for constrained optimization problems in dynamic environments.

Federica Filippini

University of Milano-Bicocca Milan, Italy

Multi-Objective Reinforcement Learning Multi-Agent System Optimization Reinforcement Learning Reward Shaping Constrained RL Multi-Agent Reinforcement Learning

Jun 16, 2026cs.AI

Treatment Response Optimized Clinical Decision Support AI System via Digital Twin Simulation

Clinical decision support AI systems (CDSASs) must adapt to evolving patient conditions in real-time while adhering to strict safety constraints. We present an online adaptive framework that integrates Treatment Effect (TE) estimation to quantify clinical benefits, a patient Digital Twin (DT) to simulate treatment trajectories, and Reinforcement Learning (RL) for sequential decision-making. The AI system is initially trained on historical medical records and operates in a continuous learning loop. To ensure safety, a rule-based module monitors vital signs and blocks contraindicated treatments. Cases with strong internal model disagreement are flagged for clinician review, simulated in our experiments via a pre-trained outcome model. We validate our framework using both a synthetic clinical simulator and a real-world ovarian cancer dataset from The Cancer Genome Atlas (TCGA). In both simulated and clinical settings, our method demonstrated superior effectiveness and stability in recommending treatments compared to standard computational baselines. Furthermore, the AI system maintains low latency and requires expert consultation for only a minority of cases in our experimental validation, demonstrating its potential as a safe, clinician-supervised tool for personalized medicine that continuously improves through practical use.

Xinyu Qin, Anil K. Sood, Ruiheng Yu +3

Department of Biomedical Engineering, University of Houston, USA · Department of Gynecologic Oncology and Reproductive Medicine, The University of Texas MD Anderson Cancer Center, USA · Department of Health Systems & Population Health Sciences, University of Houston, USA

Clinical Decision Support Constrained RL Human-in-the-Loop AI Digital Twins

Jun 15, 2026cs.RO

SafeDojo: Safe Reinforcement Learning for VLA via Interactive World Model

Safe control is a prerequisite for real-world embodied intelligence, for which safe reinforcement learning has emerged as a promising paradigm. However, existing safe reinforcement learning methods either require costly real-world exploration or depend on hand-crafted safety functions. Neither scales to vision-language-action models deployed in open-world physical environments. We propose SafeDojo, the first model-based safe reinforcement learning framework for vision-language-action policies designed to learn safe actions through world model-based imagination. Specifically, SafeDojo performs online reinforcement learning on top of an interactive video world model. The world model generates action-conditioned future predictions, from which a tailored ResNet success classifier estimates per-step task progress from imagined frames and a lightweight safety head predicts per-step safety costs from latent context together with the proposed action chunk, enabling simultaneous assessment of task execution and trajectory safety. The decoupled task-reward and safety-cost signals are balanced through a Lagrangian-based constrained GRPO objective, enabling coordinated improvement of task success and safety under explicit constraints. On SafeLIBERO, SafeDojo achieves the best aggregate task success, safe success, and execution efficiency among inference-time safety, model-free RL, and model-based RL baselines, with the best average safe-success rate on both levels and an 8.25 percentage-point improvement over the strongest baseline on Level I. Real-world Franka deployment further shows the best average task and safe-success rates across five tasks. Our results position world model-based safe reinforcement learning as a scalable and generalizable path toward safe embodied intelligence.

Kai Tang, Peidong Jia, Zhong Chu +15

State Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University · Beijing Innovation Center of Humanoid Robotics · Nanyang Technological University +2

Constrained RL Robot Safety Vision-Language-Action Models Safe RL Embodied World Models Model-Based RL

Jun 13, 2026eess.SY

Hamilton-Jacobi Reachability-Based Safe Reinforcement Learning for Emergency Collision Avoidance

Emergency collision avoidance under extreme driving conditions demands safety-critical control that accounts for both obstacle proximity and vehicle dynamic stability over a future time horizon, yet existing methods often rely on instantaneous or local safety evaluations. This paper proposes a safe reinforcement learning framework guided by a Hamilton-Jacobi (HJ) reachability based motion safety set that provides forward-looking safety supervision for constrained policy optimization. Specifically, a unified signed safety function is formulated by combining geometric collision margins and chassis stability limits, and is then extended through reachability analysis into a finite-horizon motion safety set that characterizes whether safety can be maintained under future vehicle state evolution. To enable practical computation, the motion safety set is approximated from offline extreme driving data, mitigating the computational burden of grid-based HJ solvers. The learned motion safety set is then embedded as a continuous safety cost into a constrained Markov decision process, and a PID-Lagrangian policy optimization scheme is employed to adaptively regulate the Lagrange multiplier for safety constraint enforcement. Simulation and real-vehicle experiments on low-adhesion obstacle-avoidance scenarios demonstrate that the proposed method achieves higher goal-reaching rates, produces smoother avoidance maneuvers, and maintains larger unified safety margins than baseline methods.

Yuhong Jiang, Shiyue Zhao, Junzhi Zhang +4

School of Vehicle and Mobility, Tsinghua University, Beijing, China · State Key Laboratory of Intelligent Green Vehicle and Mobility, Tsinghua University, Beijing, China

Collision Avoidance Reachability Analysis Constrained RL RL for Autonomous Driving Safe RL

Jun 12, 2026cs.AI

CSPO: Constraint-Sensitive Policy Optimization for Safe Reinforcement Learning

Safe reinforcement learning (Safe RL) aims to maximize expected return while satisfying safety constraints, typically modeled as Constrained Markov Decision Processes (CMDPs). While primal-dual methods scale well to deep RL, they often suffer from delayed constraint correction, leading to oscillatory behavior and prolonged safety violations. In this paper, we propose Constraint-Sensitive Policy Optimization (CSPO), a first-order primal-dual method that incorporates local constraint sensitivity into policy updates. CSPO augments the primal objective with a constraint-sensitive correction derived from the shortest signed distance to the safety boundary, enabling smarter recovery steps back to safety, compensating for delayed Lagrange multiplier updates, reducing oscillations near the boundary, and preserving the KKT solutions of the original constrained problem. Experiments on navigation and locomotion benchmarks demonstrate that CSPO achieves faster safety recovery and high reward preservation, resulting in higher constrained returns compared to state-of-the-art primal-dual and penalty-based methods

Ayoub Belouadah, Sylvain Kubler, Yves Le Traon

SnT, University of Luxembourg

Constrained RL Safe RL

Jun 12, 2026cs.LG

Utility-Constrained Policy Optimization

Constrained MDPs (CMDPs) are a widely adopted framework for incorporating safety into RL agents; however, the framework does not support risk-sensitive constraints. This can be problematic: For example, CMDPs allow for optimal solutions that, in order to satisfy the risk-neutral constraints, mix infrequent catastrophic behaviors and frequent, overly conservative ones. Moreover, prior empirical results suggest that enforcing stricter, risk-sensitive constraints can improve performance even under risk-neutral evaluation. The natural framework to incorporate risk-sensitive constraints is utility-constrained MDPs (UCMDPs), but no practical solutions for this problem existed. In this work, we introduce a simple yet powerful methodology for UCMDPs and constrained RL. Besides allowing for risk-sensitive constraints, our framework does not require us to fix constraint limits in advance of training the agent, provided that a sensible range is known. This increases policy flexibility and, in practice, allows for adjustments to these limits at no extra training cost. Besides benefiting from the generality of the framework, our agent shows strong performance in practice, consistently matching or outperforming existing baselines in several Safety Gymnasium benchmark tasks.

Mehrdad Moghimi, Bernardo Avila Pires

York University Toronto, Canada · Google DeepMind London, United Kingdom

Constrained RL Risk-Sensitive RL Safe RL

Jun 11, 2026math.OC

Distribution-Agnostic Robust Trajectory Optimization via Chance-Constrained Reinforcement Learning

This paper presents a distribution-agnostic robust trajectory-optimization framework based on chance-constrained reinforcement learning. The uncertainty is represented here through initial conditions and process noise, with the only requirement being that it can be sampled. A deterministic nominal trajectory is first computed offline, and reinforcement learning is then used only to robustify that baseline through a structured affine closed-loop correction law comprising a feedforward control adjustment and time-varying feedback gains. Probabilistic feasibility is enforced empirically through rollout-based upper-tail quantiles, while terminal dispersion is regulated through covariance-feasibility penalties. The framework is assessed on two materially different trajectory design problems. The flagship case study is a three-dimensional multi-impulse Earth-Mars transfer, where the learned policy is benchmarked against a recent robust trajectory-optimization reference under Gaussian uncertainty and then evaluated under bounded uniform uncertainty and under process disturbances not seen during training. The second case study is a stochastic atmospheric pinpoint rocket landing problem, used to assess portability to a short-horizon continuous-thrust setting with drag, mass depletion, and glide-slope constraints. The results show that the proposed framework can remain competitive in upper-tail fuel cost while preserving probabilistic feasibility, and that the same robustification scaffold can be carried across heterogeneous spacecraft trajectory planning problems without redesign of its core stochastic-control structure.

Yashdeep Chaudhary, Roberto Armellin, Harry Holt +1

Faculty of Engineering, Waipapa Taumata Rau – University of Auckland, 20 Symonds Street, Auckland, 1010, New Zealand · Advanced Concepts Team, European Space Research and Technology Centre, ESA, Noordwijk, 2201 AZ, The Netherlands · Department of Industrial Engineering, Alma Mater Studiorum - Università di Bologna, Via Zamboni 33, Bologna, 40126, Italy

Trajectory Optimization Constrained RL Robust RL

Jun 9, 2026cs.LG

Seeing Before Colliding: Anticipatory Safe RL with Frozen Vision-Language Models

The cost signal that constrained-RL algorithms optimize against is almost always reactive: the simulator emits a non-zero cost only after a collision has begun, and the Lagrange multiplier of PPO-Lagrangian grows only after the episode budget has been exceeded. At race speeds, where collisions are instantaneous and irreversible, any safety mechanism that waits for cost to accumulate is structurally too late. We present VLM-Safe-RL, a framework that integrates a frozen vision-language model into the CMDP Lagrangian update as an anticipatory cost term. The framework comprises four contributions: (i) Decoupled Dual-Path CLIP, independent reward/cost paths that respect the CMDP's factorization; (ii) VLM-Lagrange, an augmented multiplier update that incorporates a per-step VLM cost as an anticipatory term; (iii) Confidence Gating, a Bayes-optimal weight derived from a logistic noise model on the CLIP margin; and (iv) VLMPPOLag, the composed algorithm. On Safety-Gymnasium FormulaOne L2, our principal evaluation (

n{=}5

seeds,

10^{6}

steps, budget

d_{\text{lim}}{=}25

) VLMPPOLag

+

Conf is the only configuration in our default budget comparison that simultaneously retains substantive return (

J_r{\approx}40

) and holds cost within budget on a majority of seeds; the five constraint-aware baselines (PPOLag, CPO, CPPOPID, CPO-CLG, PPOLag-RND) each fail at least one requirement. The mechanism generalizes to held-out MetaDrive Medium (catastrophe rate

41\%{\to}26\%

, 95% bootstrap CI

[-26,-5]

,pp) and shows directionally consistent transfer to Bullet Safety-Gym; we report honestly where it does not (MetaDrive Easy/Hard, Qwen2-VL backbone) and trace the Hard failure to a Lagrangian-regulation pathology rather than the VLM signal itself. To our knowledge, this is the first work to use frozen VLM signals as an anticipatory cost term inside the CMDP Lagrangian update.

Samuel Tetteh, Cody Fleming

Iowa State University Ames, Iowa, USA

Constrained RL Robot Safety RL for Autonomous Driving Safe RL

Jun 9, 2026cs.AI

ReflectiChain: Epistemic Grounding in LLM-Driven World Models for Supply Chain Resilience

AI agents in supply chains face a fundamental epistemic gap: large language models (LLMs) interpret policies but lack physical grounding, while reinforcement learning (RL) optimizes flows but is semantically blind to unstructured constraints. We introduce REFLECTICHAIN, bridging this gap through a Generative Supply Chain World Model (SC-WM) - encoding heterogeneous supply networks into a 6-dim graph-latent space with physical conservation - and Double-Loop Learning that separates epistemic uncertainty (KL-trust-region-bounded policy adaptation) from aleatoric uncertainty (stochastic latent rollouts). On Semi-Sim, a 10-node semiconductor benchmark with SIR risk propagation, 6 perturbation types, and 10 policy constraint templates, REFLECTICHAIN improves Rationale Consistency Score by 33.0% (p < 0.0001, d = 2.78), maintains 82.3% operability under adversarial shocks, and exhibits anti-fragile behavior (+40.2% gain under moderate pressure). We identify three operational epistemic mechanisms - uncertainty separation, knowledge-boundary detection, and empirical Bayesian policy updating - and discuss five limitation categories.

Jia Luo

School of Foreign Languages, Huazhong University of Science and Technology, Wuhan 430074, China.

Constrained RL LLM-Guided RL Latent World Models Supply Chain Management Robust RL

Jun 8, 2026cs.CL

One Model, Multiple Goals: Adaptive Multi-Objective Learning for E-commerce Dialogue Systems

Dialogue systems in e-commerce scenarios often need to satisfy multiple objectives: accurately reasoning over user profiles (e.g., eligibility, credit limit) to ensure correct decision-making and user state interpretation, while also generating natural and faithful responses. These goals are complementary but not identical. In this work, we propose MORE, an adaptive Multi-Objective REinforcement learning framework that jointly optimizes reasoning accuracy and linguistic naturalness. Our preliminary experiments show that directly mixing rewards with diverging optimization dynamics can cause oscillations and unstable learning. Thus, instead of optimizing a single mixed reward, we treat reasoning functions as constraints that guide policy optimization. At inference time, the system directly generates responses without explicit reasoning steps, while still benefiting from reasoning-enhanced scaffold and avoiding additional inference overhead. To better balance linguistic objectives during response generation, we introduce an adaptive multi-reward mechanism that aggregates signals such as fluency and naturalness and dynamically reweighs them via gradient feedback. We evaluate MORE on two real-world dialogue systems at ByteDance and the MultiWOZ 2.2 benchmark, where it consistently outperforms strong baselines. In 14-day online experiments on ByteDance production traffic, MORE improves overall and reached conversion by 16.53% and 30.09%, while increasing user satisfaction and reducing handoff rates. Notably, in a human-machine comparison, MORE recovers about 60% of the incremental conversion lift achieved by human agents.

Mingzhe Li, Jing Xiang, Enguo Zhou +5

Multi-Objective Reinforcement Learning RL for Language Models Constrained RL Multi-Objective Language Model Alignment Conversational Agents

Jun 7, 2026cs.AI

Can the Environment Speak for Itself? $T^{2}$ -GRPO: A Turn-Trajectory Group Relative Policy Optimization for Caregiver Agents

Optimizing large language models (LLMs) for long-horizon caregiver agents requires balancing delayed task objectives with immediate environment dynamics, such as patient distress and resistance. In dementia care, this balance is especially difficult: trajectory level rewards are too sparse for turn level credit assignment, while external LLM-based evaluators are costly and can misread fragmented or indirect patient responses. To address this issue, we propose \textbf{T}urn-\textbf{T}rajectory \textbf{G}roup \textbf{R}elative \textbf{P}olicy \textbf{O}ptimization (\textbf{T

^{2}

-GRPO}), a framework that decouples caregiver RL into two normalized reward horizons and enforces safety through a binary hard veto.

T^2

-GRPO derives dense turn-level rewards directly from environment state transitions, measuring changes in patient distress and resistance from a frozen dementia patient simulator. These environment-grounded rewards are combined with trajectory-level evaluations through independent centered-rank normalization, which preserves heterogeneous reward signals and mitigates reward collapse. Extensive experiments on dementia caregivers show that T

^{2}

-GRPO outperforms competitive baselines, indicating a substantial improvement for emotionally sensitive caregiver scenarios that effectively handles immediate patient feedback, long-term care outcomes, and safety constraints.

Yutong Song, Jiang Wu, Pengfei Zhang +4

University of California, Irvine · Independent Researcher · Kennesaw State University

Reward Shaping Healthcare Group Relative Policy Optimization Constrained RL Long-Horizon LLM Agents

Jun 7, 2026cs.LG

Reformulate LLM Reinforcement Learning for Efficient Training under Black-box Discrepancy

Reinforcement Learning (RL) has emerged as a pivotal post-training paradigm, yet it frequently suffers from unpredictable sub-optimum performance or even training collapses. Recent findings attribute these failures to a hidden train-inference discrepancy (or mismatch), stemming from the disparate underlying engines and architecture. We find that the training policy can actively self-correct such a discrepancy when provided with an appropriate learning signal. Then, we further empirically identify a discrepancy tolerance region: within this region, aggressively narrowing the discrepancy can suppress policy exploration and reduce learning efficiency, whereas outside this region, reducing excessive discrepancy improves optimization consistency and raises the achievable local performance ceiling. According to such findings, we formulate this problem as a Discrepancy-Constrained Markov Decision Process (DCMDP), where reward maximization is coupled with a constraint that aligns training-Inference behavior, achieving stable dual-objective optimization. To adaptively balance performance improvement and discrepancy control, we introduce a Lagrangian relaxation mechanism that dynamically adjusts the relative weight of the two objectives according to the current degree of discrepancy violation. This enables stable dual-objective optimization: the policy is allowed to explore freely within the tolerance region, while being guided back when the discrepancy exceeds the safe boundary. Empirically, DCMDP significantly improves the performance of 8B dense model (Qwen-3-8b) and 30B Mixture-of-Expert model (Qwen-3-30bA3b), and enables a heterogeneous training paradigm, where LLMs can be optimized in high-fidelity training setup while being explicitly aligned for low-cost, resource-constrained inference deployment.

Jiashun Liu, Runze Liu, Xu Wan +3

1Hong Kong University of Science and Technology · 2Zhejiang University · 3Tianjin University

RL for Language Models Constrained RL RL Post-Training

Jun 6, 2026cs.AI

Neuro-Symbolic Injection of LTLf Constraints in Autoregressive Reinforcement Learning Policies

In this work we study offline reinforcement learning (RL) under temporally extended task constraints expressed in Linear Temporal Logic over finite traces (LTLf). Recently, transformer-based approaches such as Trajectory Transformers and Decision Transformers have been adopted to address RL as a sequence modeling problem. However, these methods optimize purely for reward and do not account for high-level temporal requirements. Here, we introduce a neurosymbolic framework that injects LTLf background knowledge into such transformer-based RL policies. Our approach compiles LTLf formulas into deterministic finite automata (DFAs) and integrates them into the learning process through a differentiable representation and a logic-based loss function. In particular, we derive differentiable satisfaction signals from DFA progression and use them as a regularization term during training. The resulting method is architecture-agnostic across different models. We evaluate the proposed framework on navigation environments with specification suites covering combinations of safety and reachability temporal properties. Experimental results show that incorporating background knowledge not only improves constraint satisfaction, but also maintains competitive return compared to vanilla baselines.

Ashkan Ansarifard, Matteo Mancanelli, Elena Umili +1

Sapienza University of Rome · University of Rome La Sapienza, Rome, Italy

Reinforcement Learning Transformer-Based RL Constrained RL Offline RL Linear Temporal Logic

Jun 3, 2026cs.LG

Policy-Conditioned Counterfactual Credit for Verifiable Reinforcement Learning of Long-Horizon Language Agents

Reinforcement learning with verifiable rewards improves reasoning and tool use, yet long-horizon language agents still learn unsupported evidence chains, belief drift, and shortcut actions that satisfy terminal checks. Existing process rewards are mostly correlational: they reward retrieval-, reflection-, or verification-like steps without estimating whether the step contributes to final verified success under a specified intervention. We propose CVT-RL, a constrained policy-gradient algorithm with dense verifiable rewards, intervention-validity gating, and a policy-conditioned counterfactual contribution (PCCC) estimator. Deletion, semantic substitution, evidence substitution, and tool-output perturbation define separate controlled interventions; continuations are sampled from a frozen reference policy, and a selection-adjusted doubly robust estimator augments the advantage. Belief control uses only prefix-observable labels, while an augmented Lagrangian constrains unsupported claims, skipped verification, tool tampering, and unsafe calls. On long-context QA, ALFWorld, ScienceWorld, and web/tool tasks, CVT-RL improves average task success from 71.8% for compute-matched non-causal RL and 75.4% for an information-matched counterfactual-process baseline to 78.9%, improves evidence F1 from 78.9 to 82.8 over the information-matched baseline, and reduces measured hacking from 7.2% to 3.9%. Independent human audit estimates 4.6% hacking for CVT-RL versus 8.1% for the information-matched baseline, and adaptive detector-evasion attacks raise hacking only to 7.1%. Stratified bootstrap and mixed-effects tests give p<0.01 after Holm correction for all primary metrics. Carefully scoped counterfactual credit, paired with validity gating, diagnostics, and verifiable constraints, provides a reproducible route toward more reliable long-horizon RL for language agents.

Renwei Meng

Ahu

Credit Assignment in RL Long-Horizon Agent Evaluation Constrained RL Reinforcement Learning with Verifiable Rewards Counterfactual Learning RL for LLM Agents

Jun 3, 2026cs.RO

COP-Q: Safety-First Reinforcement Learning for Robot Control via Cholesky-Ordered Projection

Safe robot control requires maximizing return while satisfying safety constraints. In off-policy safe reinforcement learning, reward and safety Q-values are commonly learned by separate critic ensembles, with uncertainty handled independently for each objective. This objective-wise treatment neglects inter-objective correlation and can lead to overly conservative value estimates, thereby reducing sample efficiency. To address this issue, we propose Cholesky-Ordered Projection Q-learning (COP-Q), a safety-first method that incorporates inter-objective covariance into vector-valued Q-value estimation. COP-Q constructs a generalized confidence bound in the joint Q-value space and uses Cholesky factorization to encode objective priority in a sequential form. This preserves conservatism on safety while adaptively reducing excessive conservatism on the reward objective. The resulting estimate is used in both temporal-difference target computation and actor optimization. COP-Q incurs minimal computational overhead and is readily compatible with most existing deep Q-learning frameworks. Experiments on robot locomotion in Brax and safe navigation in Safety-Gymnasium, covering both hard- and soft-safety settings, demonstrate that COP-Q achieves strong safety performance together with competitive or improved sample efficiency relative to representative baselines.

Guopeng Li, Moritz A. Zanger, Matthijs T. J. Spaan +1

Department of Cognitive Robotics, Delft University of Technology · School of Transportation, Southeast University, China · Department of Intelligent Systems, Delft University of Technology, the Netherlands

Robotic Control Multi-Objective Reinforcement Learning Deep Q-Learning Constrained RL Robot Safety Safe RL

Jun 2, 2026cs.RO

ConTrack: Constrained Hand Motion Tracking with Adaptive Trade-off Control

Human demonstrations provide strong priors for robot manipulation, yet it is non-trivial to transfer them to execute on real robots due to the kinematic gap. In dexterous manipulation, it remains challenging to track long-horizon, contact-rich sequences even in simulators: a reference-tracking policy must keep objects on their target trajectories while preserving demonstrated joint motion and contact timing. Existing approaches often rely on hand-crafted reward tuning that require per-sequence tuning and break under limited interaction budgets. We introduce ConTrack, a reinforcement learning (RL) framework that scales with tracking data. ConTrack treats object tracking as a constraint and allocates remaining control authority to motion fidelity, which allows it to adapt task--style trade-offs online using a dual-variable update. In addition, ConTrack also stabilizes long-horizon learning with an adaptive mid-trajectory reset library that reuses policy-reachable simulator states. Our qualitative and quantitative results in simulation tracking and real robot demonstrate that ConTrack improves success and object pose accuracy significantly over prior arts while preserving joint and contact fidelity. Website: https://www.lyt0112.com/projects/ConTrack.

Yutong Liang, Quanquan Peng, Ri-Zhao Qiu +1

University of California San Diego

Robot Trajectory Tracking Long-Horizon Robotic Manipulation Contact-Rich Robotic Manipulation Constrained RL Robotic Manipulation Robot Imitation Learning

Jun 1, 2026cs.AI

Coordination Graphs for Constrained Multi-Agent Reinforcement Learning

Constrained Multi-agent reinforcement learning (CMARL) faces two intertwined challenges: the joint action space grows exponentially with the number of agents, and additional requirements couple agents in ways that reward structure alone does not capture. We introduce Coordination Graphs for Constrained Multi-Agent Reinforcement Learning (CG-CMARL), a framework that addresses both challenges by combining coordination graphs with Lagrangian duality. The system decomposes the joint problem into pairwise regions, each served by a set of shared Q-functions, one for the primary objective and one for each of the constraints, so that the number of learned models is independent of the number of agents. At execution time, Max-Sum message passing coordinates actions across the factor graph, while a Lagrangian multiplier controls the objective--constraint tradeoff, allowing a single trained model to trace a Pareto front without retraining. We provide convergence guarantees under mild conditions, together with a compositional error bound that decomposes into separate interpretable sources, each traceable to a specific design choice and independently controllable. Experiments on cooperative navigation tasks (where teams of up to 10 agents must coordinate to reach target positions while satisfying pairwise constraints) show that our method produces Pareto fronts dominating established baselines trained at fixed reward-shaping ratios, while scaling to team sizes where centralized approaches become intractable.

Santiago Amaya-Corredor, Miguel Calvo-Fullana, Anders Jonsson

Department of Engineering, Universitat Pompeu Fabra, Barcelona, Spain

Multi-Objective Reinforcement Learning Multi-Agent System Optimization Constrained RL Constrained Optimization Cooperative MARL

May 31, 2026cs.LG

Lagrangian Perturbation Diffusion Steering: Latent Reinforcement Learning for Generative Policies

Behavior cloning with high-capacity generative policies achieves strong imitation performance, but is often limited by demonstration coverage and distribution shift. Direct reinforcement learning fine-tuning can improve performance, but updating large action decoders is frequently unstable and sample inefficient. We propose Lagrangian Perturbation Diffusion Steering (LP-DS), a lightweight adaptation method that improves a frozen generative policy by learning a compact noise-space perturbation before decoding. LP-DS optimizes this perturbation with a Lagrangian trust-region objective, improving downstream value while constraining deviation from the latent prior. Across RoboMimic manipulation, OpenAI Gym locomotion, and Adroit dexterous manipulation benchmarks, LP-DS improves sample efficiency, success, and return while maintaining higher action-space entropy than unconstrained noise-space steering, with return improvements of up to 25% over prior baselines. Additional evaluations with flow-matching backbones, a large vision-language-action model, and physical Franka deployment show that LP-DS is not limited to compact diffusion policies or simulated benchmarks. Project page: https://sites.google.com/view/lp-ds/home.

Hikmet Simsir, Ozgur S. Oguz

Department of Computer Engineering, Bilkent University, Ankara, Türkiye.

Robot Policy Adaptation Reinforcement Learning Constrained RL Robotic RL Diffusion Policy RL for Diffusion Models+2

May 29, 2026cs.LG

Detector-Evasive LLM Paraphrasing via Constrained Policy Optimization

AI-text detectors are vulnerable to paraphrasing and detector-guided paraphrasing attacks, but existing detector-evasion methods often lack precise control over semantic preservation. In particular, optimizing directly for detector evasion can degrade fine-grained semantics, whereas scalarized reward designs provide only indirect, weight-sensitive control over the evasion-semantics trade-off. We address this limitation by formulating detector-evasive LLM paraphrasing as a Constrained Markov Decision Process, where detector evasion is the primary objective and semantic preservation is enforced as an explicit constraint. We propose Detector Evasion Policy Optimization (DEPO), a Lagrangian primal-dual reinforcement learning algorithm with a novel GRPO-style group-based policy update. DEPO adaptively balances semantic preservation and detector evasion during training, enabling the policy to improve attack success within a prescribed semantic-preservation region. Experiments on MAGE, M4, RAID, and peer-review datasets, evaluated against MAGE, RoBERTa, RADAR, Binoculars, and Fast-DetectGPT detectors, show that DEPO achieves strong detector evasion while precisely satisfying the semantic preservation constraint. DEPO also exhibits cross-domain, cross-detector, and prompt-level robustness.

Mingyi Wang, Zhuoer Shen, Yuheng Bu +1

School of ECEE, Arizona State University · Department of Computer Science, University of California, Santa Barbara

Constrained RL Text Generation Adversarial Text Attacks

May 29, 2026cs.RO

Constrained Whole-Body Tracking for Humanoid Robots

Recent advances in reinforcement learning (RL) have demonstrated impressive whole-body agility for humanoid robots, yet ensuring safety and satisfying constraints -- particularly those specified after training -- remains a challenge. Towards this goal, we present ConstrainedMimic, a control framework that leverages whole-body kinematics and dynamics for real-time constraint enforcement within RL tracking policies. By integrating principles from operational space control and control barrier functions (CBFs), we enable the satisfaction of arbitrary runtime constraints on both the kinematic reference motion and the underlying dynamics. In whole-body motion-tracking and teleoperation experiments on a (simulated) Unitree G1 with a learned policy, we demonstrate collision avoidance (both with the robot body and external obstacles), joint limits, and center of mass stability constraints. By remaining consistent with the current contact mode and tracking objectives, we minimally restrict the capabilities of the policy when constraints are active. Our method is fully differentiable, runs on CPU, GPU, and TPU, and can be deployed at up to 300-500 Hz. All software will be freely available upon publication.

Daniel Morton, Pranit Mohnot, Marco Pavone

Stanford University · NVIDIA Research

Robotic Control Robot Trajectory Tracking Humanoid Robot Control Constrained RL Whole-Body Control Humanoid Motion Tracking

May 29, 2026cs.LG

Constrained Multi-Objective Reinforcement Learning with Max-Min Criterion

Multi-Objective Reinforcement Learning (MORL) extends standard RL by optimizing policies with respect to multiple, often conflicting, objectives. While max-min MORL has emerged as an effective approach for promoting fairness, its applicability remains limited, particularly when constraints must be incorporated. In this paper, we propose a MORL framework that integrates the max-min criterion with explicit constraint satisfaction. We establish a theoretical foundation for the proposed framework and validate the resulting algorithm through convergence analysis and experiments in tabular settings. We further demonstrate the practical relevance of our approach in simulated building thermal control, multi-objective locomotion control, and greenhouse-gas-emission-aware traffic management. Across these domains, our method effectively balances fairness and constraint satisfaction in multi-objective decision-making.

Giseung Park, Hyunyoung Nam, Woohyeon Byeon +2

Robotics Institute, University of Toronto, 55 St George St, Toronto, Canada. · School of Electical Engineering, KAIST, Daejeon 34141, Republic of Korea. · Faculty of Engineering, Bar-Ilan University, Ramat Gan 52900, Israel.

Multi-Objective Reinforcement Learning Reinforcement Learning Constrained RL

May 26, 2026cs.LG

Transferable Reinforcement Learning via Probabilistic Latent Embeddings and Dynamic Policy Adaptation for Sim-to-Real Deployment

Due to limited resources and public safety concerns, deep reinforcement learning (RL) agents for many cyber-physical systems (e.g., autonomous vehicles) are first trained in simulators. However, when deployed in real world environments, they often suffer from performance degradation or safety violations because of the inevitable Sim2Real gap. Existing zero-shot approaches, such as robust safe RL and domain randomization, mitigate this issue but typically at the cost of degraded performance or residual safety risks when experiencing unmodeled system dynamics. To address these limitations, we propose a novel reinforcement learning framework that enables safe and efficient policy transfer via probabilistic latent embeddings and dynamic policy adaptation. We consider a family of Constrained Markov Decision Processes (CMDPs) under different environment contexts. By leveraging latent context variable in meta-RL, the proposed framework infers the latent representation of the environment from simulated experiences. Furthermore, it incorporates a distributional RL formulation, which allows risk levels of the deployed policy to be adjusted dynamically, based on the estimation accuracy of the latent context variable. This strategy promotes safety at the early deployment stage and improves efficiency through fast policy adaptation under the Sim2Real gap.

Gengyue Han, Yiheng Feng

Lyles School of Civil and Construction Engineering, Purdue University, West Lafayette, USA · Elmore Family School of Electrical and Computer Engineering, Purdue University, West Lafayette, USA

Reinforcement Learning Sim-to-Real Transfer Constrained RL Meta-Reinforcement Learning Safe RL

May 26, 2026eess.SY

Breaking the Epistemic Trap: Active Perception Under Compound Uncertainty

Deploying reinforcement learning in safety critical domains, from autonomous vehicles to medical decision support, is constrained by failures arising when systems encounter unfamiliar conditions. We argue that the fundamental bottleneck is not individual challenges like changing dynamics or incomplete observations, but their synergistic interaction, which we term the Epistemic Trap: agents cannot estimate their state without knowing system dynamics, nor learn dynamics without accurate state information. Proof-of-concept experiments in simulated locomotion reveal that combining these uncertainties causes failures far worse than either challenge alone, a 77% observed degradation against the 46% additive prediction, demonstrating that compounding failure modes can emerge and, when they do, far exceed what additive reasoning would predict. Conventional approaches typically adopt a passive epistemic stance that cannot resolve this coupled uncertainty. We propose reframing safety as an information problem. We introduce an Adaptive Safety Architecture built around three contributions. First, the Compound Uncertainty Coefficient (

κ

), a mutual-information based metric that quantifies how tightly state and dynamics uncertainties are coupled. Second, information-seeking policies governed by a MaxInfoRL objective that actively probe system dynamics rather than waiting for the environment to reveal itself passively. Third, regime adaptive safety constraints that tighten automatically as epistemic coupling rises. Together, these constitute a paradigm shift from passive robustness to active perception, offering a principled path toward decision making systems that operate under uncertainty, recognize their own ignorance, and act strategically to resolve it.

Chayan Banerjee, Ethan Goan

School of Electrical Engineering and Robotics, Queensland University of Technology, Brisbane, Australia.

Active Perception Constrained RL RL Exploration Safe RL Epistemic Uncertainty