-- This paper introduces the Cable Robot Simulation and Control (CaRoSaC) Framework, which integrates a realistic simulation environment with a model-free reinforcement learning control methodology for suspended Cable-Driven Parallel Robots (CDPRs), accounting for the effects of cable sag. Our approach seeks to bridge the knowledge gap of the intricacies of CDPRs due to aspects such as cable sag and precision control necessities, which are missing in existing research and often overlooked in traditional models, by establishing a simulation platform that captures the real-world behaviors of CDPRs, including the impacts of cable sag. The framework offers researchers and developers a tool to further develop estimation and control strategies within the simulation for understanding and predicting the performance nuances, especially in complex operations where cable sag can be significant. Using this simulation framework, we train a model-free control policy rooted in Reinforcement Learning (RL). This approach is chosen for its capability to adaptively learn from the complex dynamics of CDPRs. The policy is trained to discern optimal cable control inputs, ensuring precise end-effector positioning. Unlike traditional feedback-based control methods, our RL control policy focuses on kinematic control and addresses the cable sag issues without being tethered to predefined mathematical models. We also demonstrate that our RL-based controller, coupled with the flexible cable simulation, significantly outperforms the classical kinematics approach, particularly in dynamic conditions and near the boundary regions of the workspace. The combined strength of the described simulation and control approach offers an effective solution in manipulating suspended CDPRs even at workspace boundary conditions where traditional approach fails, as proven from our experiments, ensuring that CDPRs function optimally in various applications while accounting for the often neglected but critical factor of cable sag. CDPRs have emerged as a powerful subset of parallel manipulators, offering enhanced flexibility due to the replacement of rigid links with flexible cables.

This article presents an entirely data-driven approach for autonomous control of redundant manipulators with hydraulic actuation. The approach only requires minimal system information, which is inherited from a simulation model. The non-linear hydraulic actuation dynamics are modeled using actuator networks from the data gathered during the manual operation of the manipulator to effectively emulate the real system in a simulation environment. A neural network control policy for autonomous control, based on end-effector (EE) position tracking is then learned using Reinforcement Learning (RL) with Ornstein-Uhlenbeck process noise (OUNoise) for efficient exploration. The RL agent also receives feedback based on supervised learning of the forward kinematics which facilitates selecting the best suitable action from exploration. The control policy directly provides the joint variables as outputs based on provided target EE position while taking into account the system dynamics. The joint variables are then mapped to the hydraulic valve commands, which are then fed to the system without further modifications. The proposed approach is implemented on a scaled hydraulic forwarder crane with three revolute and one prismatic joint to track the desired position of the EE in 3-Dimensional (3D) space. With the emulated dynamics and extensive learning in simulation, the results demonstrate the feasibility of deploying the learned controller directly on the real system.

We present a novel reinforcement learning (RL) approach for solving the classical 2-level atom non-LTE radiative transfer problem by framing it as a control task in which an RL agent learns a depth-dependent source function $S(τ)$ that self-consistently satisfies the equation of statistical equilibrium (SE). The agent's policy is optimized entirely via reward-based interactions with a radiative transfer engine, without explicit knowledge of the ground truth. This method bypasses the need for constructing approximate lambda operators ($Λ^*$) common in accelerated iterative schemes. Additionally, it requires no extensive precomputed labeled datasets to extract a supervisory signal, and avoids backpropagating gradients through the complex RT solver itself. Finally, we show through experiment that a simple feedforward neural network trained greedily cannot solve for SE, possibly due to the moving target nature of the problem. Our $Λ^*-\text{Free}$ method offers potential advantages for complex scenarios (e.g., atmospheres with enhanced velocity fields, multi-dimensional geometries, or complex microphysics) where $Λ^*$ construction or solver differentiability is challenging. Additionally, the agent can be incentivized to find more efficient policies by manipulating the discount factor, leading to a reprioritization of immediate rewards. If demonstrated to generalize past its training data, this RL framework could serve as an alternative or accelerated formalism to achieve SE. To the best of our knowledge, this study represents the first application of reinforcement learning in solar physics that directly solves for a fundamental physical constraint.

Transformer-based architectures have recently propelled advances in sequence modeling across domains, but their application to the hydrophobic-hydrophilic (H-P) model for protein folding remains relatively unexplored. In this work, we adapt a Deep Q-Network (DQN) integrated with attention mechanisms (Transformers) to address the 3D H-P protein folding problem. Our system formulates folding decisions as a self-avoiding walk in a reinforced environment, and employs a specialized reward function based on favorable hydrophobic interactions. T o improve performance, the method incorporates validity check including symmetry-breaking constraints, dueling and double Q-learning, and prioritized replay to focus learning on critical transitions. Experimental evaluations on standard benchmark sequences demonstrate that our approach achieves several known best solutions for shorter sequences, and obtains near-optimal results for longer chains. This study underscores the promise of attention-based reinforcement learning for protein folding, and created a prototype of Transformer-based Q-network structure for 3-dimensional lattice models. 1 1 Introduction H-P model has been considered as a simplified model for protein structure prediction. However, optimizing the structure of H-P model still requires efficient algorithms due to the large solution space. Determining the optimal structure of proteins under the hydrophobic-hydrophilic (HP) model has been rigorously shown to be NP-complete ( 1), highlighting the necessity for powerful heuristic or approximation methods in lieu of brute-force searches. Among heuristic approaches, Monte Carlo simulations are particularly popular and exhibit a wide range of implementations ( 2) ( 3).

Abstract: Accurately identifying the parameters of electrochemical models of li-ion battery (LiB) cells is a critical task for enhancing the fidelity and predictive ability. Traditional parameter identification methods often require extensive data collection experiments and lack adaptability in dynamic environments. This paper describes a Reinforcement Learning (RL) based approach that dynamically tailors the current profile applied to a LiB cell to optimize the parameters identifiability of the electrochemical model. The proposed framework is implemented in real-time using a Hardware-in-the-Loop (HIL) setup, which serves as a reliable testbed for evaluating the RL-based design strategy. The HIL validation confirms that the RL-based experimental design outperforms conventional test protocols used for parameter identification in terms of both reducing the modeling errors on a verification test and minimizing the duration of the experiment used for parameter identification.

Unlike prior approaches that rely heavily on reward engineering, human demonstrations, motion capture, or expensive pairwise preference labels, LAPP leverages large language models (LLMs) to automatically generate preference labels from raw state-action trajectories collected during reinforcement learning (RL). These labels are used to train an online preference predictor, which in turn guides the policy optimization process toward satisfying high-level behavioral specifications provided by humans. Our key technical contribution is the integration of LLMs into the RL feedback loop through trajectory-level preference prediction, enabling robots to acquire complex skills including subtle control over gait patterns and rhythmic timing. We evaluate LAPP on a diverse set of quadruped locomotion and dexterous manipulation tasks and show that it achieves efficient learning, higher final performance, faster adaptation, and precise control of high-level behaviors. Notably, LAPP enables robots to master highly dynamic and expressive tasks such as quadruped backflips, which remain out of reach for standard LLM-generated or handcrafted rewards. Our results highlight LAPP as a promising direction for scalable preference-driven robot learning.Figure 1: Large Language Model-Assisted Preference Prediction (LAPP) takes in language behavior instructions and generates preference feedback to guide reinforcement learning training from raw state-action robot trajectories.

Tasks for multi-robot systems often require the robots to collaborate and complete a team goal while maintaining safety. This problem is usually formalized as a constrained Markov decision process (CMDP), which targets minimizing a global cost and bringing the mean of constraint violation below a user-defined threshold. Inspired by real-world robotic applications, we define safety as zero constraint violation. While many safe multi-agent reinforcement learning (MARL) algorithms have been proposed to solve CMDPs, these algorithms suffer from unstable training in this setting. To tackle this, we use the epigraph form for constrained optimization to improve training stability and prove that the centralized epigraph form problem can be solved in a distributed fashion by each agent. This results in a novel centralized training distributed execution MARL algorithm named Def-MARL. Simulation experiments on 8 different tasks across 2 different simulators show that Def-MARL achieves the best overall performance, satisfies safety constraints, and maintains stable training. Real-world hardware experiments on Crazyflie quadcopters demonstrate the ability of Def-MARL to safely coordinate agents to complete complex collaborative tasks compared to other methods.

We introduce AlphaGrad, a memory-efficient, conditionally stateless optimizer addressing the memory overhead and hyperparameter complexity of adaptive methods like Adam. AlphaGrad enforces scale invariance via tensor-wise L2 gradient normalization followed by a smooth hyperbolic tangent transformation, $g' = \tanh(\alpha \cdot \tilde{g})$, controlled by a single steepness parameter $\alpha$. Our contributions include: (1) the AlphaGrad algorithm formulation; (2) a formal non-convex convergence analysis guaranteeing stationarity; (3) extensive empirical evaluation on diverse RL benchmarks (DQN, TD3, PPO). Compared to Adam, AlphaGrad demonstrates a highly context-dependent performance profile. While exhibiting instability in off-policy DQN, it provides enhanced training stability with competitive results in TD3 (requiring careful $\alpha$ tuning) and achieves substantially superior performance in on-policy PPO. These results underscore the critical importance of empirical $\alpha$ selection, revealing strong interactions between the optimizer's dynamics and the underlying RL algorithm. AlphaGrad presents a compelling alternative optimizer for memory-constrained scenarios and shows significant promise for on-policy learning regimes where its stability and efficiency advantages can be particularly impactful.

Advancing the dynamic loco-manipulation capabilities of quadruped robots in complex terrains is crucial for performing diverse tasks. Specifically, dynamic ball manipulation in rugged environments presents two key challenges. The first is coordinating distinct motion modalities to integrate terrain traversal and ball control seamlessly. The second is overcoming sparse rewards in end-to-end deep reinforcement learning, which impedes efficient policy convergence. To address these challenges, we propose a hierarchical reinforcement learning framework. A high-level policy, informed by proprioceptive data and ball position, adaptively switches between pre-trained low-level skills such as ball dribbling and rough terrain navigation. We further propose Dynamic Skill-Focused Policy Optimization to suppress gradients from inactive skills and enhance critical skill learning. Both simulation and real-world experiments validate that our methods outperform baseline approaches in dynamic ball manipulation across rugged terrains, highlighting its effectiveness in challenging environments. Videos are on our website: dribble-hrl.github.io.

As cloud computing continues to evolve, the adoption of multi-NUMA (Non-Uniform Memory Access) architecture by cloud service providers has introduced new challenges in virtual machine (VM) scheduling. To address these challenges and more accurately reflect the complexities faced by modern cloud environments, we introduce the Dynamic VM Allocation problem in Multi-NUMA PM (DVAMP). We formally define both offline and online versions of DVAMP as mixed-integer linear programming problems, providing a rigorous mathematical foundation for analysis. A tight performance bound for greedy online algorithms is derived, offering insights into the worst-case optimality gap as a function of the number of physical machines and VM lifetime variability. To address the challenges posed by DVAMP, we propose SPANE (Symmetry-Preserving Architecture for Multi-NUMA Environments), a novel deep reinforcement learning approach that exploits the problem's inherent symmetries. SPANE produces invariant results under arbitrary permutations of physical machine states, enhancing learning efficiency and solution quality. Extensive experiments conducted on the Huawei-East-1 dataset demonstrate that SPANE outperforms existing baselines, reducing average VM wait time by 45%. Our work contributes to the field of cloud resource management by providing both theoretical insights and practical solutions for VM scheduling in multi-NUMA environments, addressing a critical gap in the literature and offering improved performance for real-world cloud systems.