Institute of Robotics and Mechatronics, German Aerospace Center (DLR), Oberpfaffenhofen, Muenchener Strasse 20, 82234, Wessling, Germany Abstract: In this work, we present a novel approach to augment a model-based control method with a reinforcement learning (RL) agent and demonstrate a swing-up maneuver with a suspended aerial manipulation platform. These platforms are targeted towards a wide range of applications on construction sites involving cranes, with swing-up maneuvers allowing it to perch at a given location, inaccessible with purely the thrust force of the platform. Our proposed approach is based on a hierarchical control framework, which allows different tasks to be executed according to their assigned priorities. An RL agent is then subsequently utilized to adjust the reference set-point of the lower-priority tasks to perform the swing-up maneuver, which is confined in the nullspace of the higher-priority tasks, such as maintaining a specific orientation and position of the end-effector. Our approach is validated using extensive numerical simulation studies.

Owing to its rapid progress and broad application prospects, few-shot action recognition has attracted considerable interest. However, current methods are predominantly based on limited single-modal data, which does not fully exploit the potential of multimodal information. This paper presents a novel framework that actively identifies reliable modalities for each sample using task-specific contextual cues, thus significantly improving recognition performance. Our framework integrates an Active Sample Inference (ASI) module, which utilizes active inference to predict reliable modalities based on posterior distributions and subsequently organizes them accordingly. Unlike reinforcement learning, active inference replaces rewards with evidence-based preferences, making more stable predictions. Additionally, we introduce an active mutual distillation module that enhances the representation learning of less reliable modalities by transferring knowledge from more reliable ones. Adaptive multimodal inference is employed during the meta-test to assign higher weights to reliable modalities. Extensive experiments across multiple benchmarks demonstrate that our method significantly outperforms existing approaches.

This paper develops a novel multi-agent reinforcement learning (MARL) framework for reinsurance treaty bidding, addressing long-standing inefficiencies in traditional broker-mediated placement processes. We pose the core research question: Can autonomous, learning-based bidding systems improve risk transfer efficiency and outperform conventional pricing approaches in reinsurance markets? In our model, each reinsurer is represented by an adaptive agent that iteratively refines its bidding strategy within a competitive, partially observable environment. The simulation explicitly incorporates institutional frictions including broker intermediation, incumbent advantages, last-look privileges, and asymmetric access to underwriting information. Empirical analysis demonstrates that MARL agents achieve up to 15% higher underwriting profit, 20% lower tail risk (CVaR), and over 25% improvement in Sharpe ratios relative to actuarial and heuristic baselines. Sensitivity tests confirm robustness across hyperparameter settings, and stress testing reveals strong resilience under simulated catastrophe shocks and capital constraints. These findings suggest that MARL offers a viable path toward more transparent, adaptive, and risk-sensitive reinsurance markets. The proposed framework contributes to emerging literature at the intersection of algorithmic market design, strategic bidding, and AI-enabled financial decision-making.

This work presents a n approach to the inverse design of scattering systems by modifying the transmission matrix u sing reinforcement learning . We utilize Proximal Policy Optimization to navigate the highly non - convex landscape of the object function to achieve three types of transmission matri ces: (1) F ixed - ratio power conversion and z ero - transmission mode in r ank - 1 matri ces, (2) exceptional points with degenerate eigenvalues and unidirectional mode conversion, and (3) uniform channel participation is enforced when transmission eigenvalues are degenerate . Engineering wave propagation is a fast - moving domain. S ingularit ies of the scattering matrix (SM), or sub - SM, such as the transmission matrix (TM) or reflection matrix (RM) encode the scattering behavior of a n open system and can be exploited in sensing, switching, lasing and energy deposition [1,2] . Open system s can be described by effective non - Hermitian Hamiltonians, and their resonance frequencies corresponds to poles of SM. Frequency points at which the response vanishes are described by zeros, which are also usually complex value. Incident radiation is completely absorbed when a zero of the SM is brought to the real axis. Such coherent perfect absorption (CPA) is the time reversal of an outgoing wave at the lasing threshold [4] .

Recent advances in Trajectory Optimization (TO) models have achieved remarkable success in offline reinforcement learning. However, their vulnerabilities against backdoor attacks are poorly understood. We find that existing backdoor attacks in reinforcement learning are based on reward manipulation, which are largely ineffective against the TO model due to its inherent sequence modeling nature. Moreover, the complexities introduced by high-dimensional action spaces further compound the challenge of action manipulation. To address these gaps, we propose TrojanTO, the first action-level backdoor attack against TO models. TrojanTO employs alternating training to enhance the connection between triggers and target actions for attack effectiveness. To improve attack stealth, it utilizes precise poisoning via trajectory filtering for normal performance and batch poisoning for trigger consistency. Extensive evaluations demonstrate that TrojanTO effectively implants backdoor attacks across diverse tasks and attack objectives with a low attack budget (0.3\% of trajectories). Furthermore, TrojanTO exhibits broad applicability to DT, GDT, and DC, underscoring its scalability across diverse TO model architectures.

The open radio access network (O-RAN) architecture introduces RAN intelligent controllers (RICs) to facilitate the management and optimization of the disaggregated RAN. Reinforcement learning (RL) and its advanced form, deep RL (DRL), are increasingly employed for designing intelligent controllers, or xApps, to be deployed in the near-real time (near-RT) RIC. These models often encounter local optima, which raise concerns about their reliability for RAN intelligent control. We therefore introduce Federated O-RAN enabled Neuroevolution (NE)-enhanced DRL (F-ONRL) that deploys an NE-based optimizer xApp in parallel to the RAN controller xApps. This NE-DRL xApp framework enables effective exploration and exploitation in the near-RT RIC without disrupting RAN operations. We implement the NE xApp along with a DRL xApp and deploy them on Open AI Cellular (OAIC) platform and present numerical results that demonstrate the improved robustness of xApps while effectively balancing the additional computational load.

We present \textbf{FlowRL}, a novel framework for online reinforcement learning that integrates flow-based policy representation with Wasserstein-2-regularized optimization. We argue that in addition to training signals, enhancing the expressiveness of the policy class is crucial for the performance gains in RL. Flow-based generative models offer such potential, excelling at capturing complex, multimodal action distributions. However, their direct application in online RL is challenging due to a fundamental objective mismatch: standard flow training optimizes for static data imitation, while RL requires value-based policy optimization through a dynamic buffer, leading to difficult optimization landscapes. FlowRL first models policies via a state-dependent velocity field, generating actions through deterministic ODE integration from noise. We derive a constrained policy search objective that jointly maximizes Q through the flow policy while bounding the Wasserstein-2 distance to a behavior-optimal policy implicitly derived from the replay buffer. This formulation effectively aligns the flow optimization with the RL objective, enabling efficient and value-aware policy learning despite the complexity of the policy class. Empirical evaluations on DMControl and Humanoidbench demonstrate that FlowRL achieves competitive performance in online reinforcement learning benchmarks.

Due to hardware constraints and the inherent complexity of the robotic action space, achieving effective whole-body control (WBC) for adult-sized humanoid robots, such as the Unitree H1-2, remains a significant challenge. Recent studies on WBC have demonstrated promising advancements, enabling humanoid robots to perform versatile motions by learning from extensive human data [1, 2, 3, 4, 5]. Based on different task settings, WBC methodologies can be broadly categorized into three types: root velocity tracking [6], kinematic position tracking [1, 3], and local joint angle tracking [6, 2, 4]. Root velocity tracking emphasizes coarse-grained control, where the robot tracks a given velocity without relying on a specific reference pose. In contrast, kinematic position and local joint angle tracking focus on accurately reproducing a given trajectory of reference poses, which can be regarded as fine-grained control for humanoids.

The collaboration and interaction of multiple robots have become integral aspects of smart manufacturing. Effective planning and management play a crucial role in achieving energy savings and minimising overall costs. This paper addresses the real-time Dynamic Multiple Sources to Single Destination (DMS-SD) navigation problem, particularly with a material distribution case for multiple intelligent robots in smart manufacturing. Enumerated solutions, such as in \cite{xiao2022efficient}, tackle the problem by generating as many optimal or near-optimal solutions as possible but do not learn patterns from the previous experience, whereas the method in \cite{xiao2023collaborative} only uses limited information from the earlier trajectories. Consequently, these methods may take a considerable amount of time to compute results on large maps, rendering real-time operations impractical. To overcome this challenge, we propose a lightweight Deep Reinforcement Learning (DRL) method to address the DMS-SD problem. The proposed DRL method can be efficiently trained and rapidly converges to the optimal solution using the designed target-guided reward function. A well-trained DRL model significantly reduces the computation time for the next movement to a millisecond level, which improves the time up to 100 times in our experiments compared to the enumerated solutions. Moreover, the trained DRL model can be easily deployed on lightweight devices in smart manufacturing, such as Internet of Things devices and mobile phones, which only require limited computational resources.

In reinforcement learning (RL), value-based algorithms learn to associate each observation with the states and rewards that are likely to be reached from it. We observe that many self-supervised image pre-training methods bear similarity to this formulation: learning features that associate crops of images with those of nearby views, e.g., by taking a different crop or color augmentation. In this paper, we complete this analogy and explore a method that directly casts pre-training on unlabeled image data like web crawls and video frames as an RL problem. We train a general value function in a dynamical system where an agent transforms an image by changing the view or adding image augmentations. Learning in this way resembles crop-consistency self-supervision, but through the reward function, offers a simple lever to shape feature learning using curated images or weakly labeled captions when they exist. Our experiments demonstrate improved representations when training on unlabeled images in the wild, including video data like EpicKitchens, scene data like COCO, and web-crawl data like CC12M.