AI Agents powered by Large Language Models are transforming the world through enormous applications. A super agent has the potential to fulfill diverse user needs, such as summarization, coding, and research, by accurately understanding user intent and leveraging the appropriate tools to solve tasks. However, to make such an agent viable for real-world deployment and accessible at scale, significant optimizations are required to ensure high efficiency and low cost. This position paper presents a design of the Super Agent System powered by the hybrid AI routers. Upon receiving a user prompt, the system first detects the intent of the user, then routes the request to specialized task agents with the necessary tools or automatically generates agentic workflows. In practice, most applications directly serve as AI assistants on edge devices such as phones and robots. As different language models vary in capability and cloud-based models often entail high computational costs, latency, and privacy concerns, we then explore the hybrid mode where the router dynamically selects between local and cloud models based on task complexity. Finally, we introduce the blueprint of an on-device super agent enhanced with cloud. With advances in multi-modality models and edge hardware, we envision that most computations can be handled locally, with cloud collaboration only as needed. Such architecture paves the way for super agents to be seamlessly integrated into everyday life in the near future.

Processing spatial data is a key component in many learning tasks for autonomous driving such as motion forecasting, multi-agent simulation, and planning. Prior works have demonstrated the value in using SE(2) invariant network architectures that consider only the relative poses between objects (e.g. other agents, scene features such as traffic lanes). However, these methods compute the relative poses for all pairs of objects explicitly, requiring quadratic memory. In this work, we propose a mechanism for SE(2) invariant scaled dot-product attention that requires linear memory relative to the number of objects in the scene. Our SE(2) invariant transformer architecture enjoys the same scaling properties that have benefited large language models in recent years. We demonstrate experimentally that our approach is practical to implement and improves performance compared to comparable non-invariant architectures.

We propose LingBench++, a linguistically-informed benchmark and reasoning framework designed to evaluate large language models (LLMs) on complex linguistic tasks inspired by the International Linguistics Olympiad (IOL). Unlike prior benchmarks that focus solely on final answer accuracy, LingBench++ provides structured reasoning traces, stepwise evaluation protocols, and rich typological metadata across over 90 low-resource and cross-cultural languages. We further develop a multi-agent architecture integrating grammatical knowledge retrieval, tool-augmented reasoning, and deliberate hypothesis testing. Through systematic comparisons of baseline and our proposed agentic models, we demonstrate that models equipped with external knowledge sources and iterative reasoning outperform single-pass approaches in both accuracy and interpretability. LingBench++ offers a comprehensive foundation for advancing linguistically grounded, culturally informed, and cognitively plausible reasoning in LLMs.

Cooperative multi-agent reinforcement learning (MARL) is typically formalised as a Decentralised Partially Observable Markov Decision Process (Dec-POMDP), where agents must reason about the environment and other agents' behaviour. In practice, current model-free MARL algorithms use simple recurrent function approximators to address the challenge of reasoning about others using partial information. In this position paper, we argue that the empirical success of these methods is not due to effective Markov signal recovery, but rather to learning simple conventions that bypass environment observations and memory. Through a targeted case study, we show that co-adapting agents can learn brittle conventions, which then fail when partnered with non-adaptive agents. Crucially, the same models can learn grounded policies when the task design necessitates it, revealing that the issue is not a fundamental limitation of the learning models but a failure of the benchmark design. Our analysis also suggests that modern MARL environments may not adequately test the core assumptions of Dec-POMDPs. We therefore advocate for new cooperative environments built upon two core principles: (1) behaviours grounded in observations and (2) memory-based reasoning about other agents, ensuring success requires genuine skill rather than fragile, co-adapted agreements.

Autonomous traffic agents (ATAs) are expected to act in ways tat are not only safe, but also aligned with stakeholder values across legal, social, and moral dimensions. In this paper, we adopt an established formal model of conflict from epistemic game theory to support the development of such agents. We focus on value conflicts-situations in which agents face competing goals rooted in value-laden situations and show how conflict analysis can inform key phases of the design process. This includes value elicitation, capability specification, explanation, and adaptive system refinement. We elaborate and apply the concept of Value-Aligned Operational Design Domains (VODDs) to structure autonomy in accordance with contextual value priorities. Our approach shifts the emphasis from solving moral dilemmas at runtime to anticipating and structuring value-sensitive behaviour during development.

The sparse identification of nonlinear dynamics (SINDy) approach can discover the governing equations of dynamical systems based on measurement data, where the dynamical model is identified as the sparse linear combination of the given basis functions. A major challenge in SINDy is the design of a library, which is a set of candidate basis functions, as the appropriate library is not trivial for many dynamical systems. To overcome this difficulty, this study proposes SINDy with library optimization mechanism (SINDy-LOM), which is a combination of the sparse regression technique and the novel learning strategy of the library. In the proposed approach, the basis functions are parametrized. The SINDy-LOM approach involves a two-layer optimization architecture: the inner-layer, in which the data-driven model is extracted as the sparse linear combination of the candidate basis functions, and the outer-layer, in which the basis functions are optimized from the viewpoint of the recursive long-term (RLT) prediction accuracy; thus, the library design is reformulated as the optimization of the parametrized basis functions. The resulting SINDy-LOM model has good interpretability and usability, as the proposed approach yields the parsimonious model. The library optimization mechanism significantly reduces user burden. The RLT perspective improves the reliability of the resulting model compared with the traditional SINDy approach that can only ensure the one-step-ahead prediction accuracy. The validity of the proposed approach is demonstrated by applying it to a diesel engine airpath system, which is a well-known complex industrial system.

To achieve full autonomous driving, a good understanding of the surrounding environment is necessary. Especially predicting the future states of other traffic participants imposes a non-trivial challenge. Current SotA-models already show promising results when trained on real datasets (e.g. Argoverse2, NuScenes). Problems arise when these models are deployed to new/unseen areas. Typically, performance drops significantly, indicating that the models lack generalization. In this work, we introduce a new Graph Neural Network (GNN) that utilizes a heterogeneous graph consisting of traffic participants and vectorized road network. Latter, is used to classify goals, i.e. endpoints of the predicted trajectories, in a multi-staged approach, leading to a better generalization to unseen scenarios. We show the effectiveness of the goal selection process via cross-dataset evaluation, i.e. training on Argoverse2 and evaluating on NuScenes.

The growing need for intelligent, adaptive, and energy-efficient autonomous systems across fields such as robotics, mobile agents (e.g., UAVs), and self-driving vehicles is driving interest in neuromorphic computing. By drawing inspiration from biological neural systems, neuromorphic approaches offer promising pathways to enhance the perception, decision-making, and responsiveness of autonomous platforms. This paper surveys recent progress in neuromorphic algorithms, specialized hardware, and cross-layer optimization strategies, with a focus on their deployment in real-world autonomous scenarios. Special attention is given to event-based dynamic vision sensors and their role in enabling fast, efficient perception. The discussion highlights new methods that improve energy efficiency, robustness, adaptability, and reliability through the integration of spiking neural networks into autonomous system architectures. We integrate perspectives from machine learning, robotics, neuroscience, and neuromorphic engineering to offer a comprehensive view of the state of the field. Finally, emerging trends and open challenges are explored, particularly in the areas of real-time decision-making, continual learning, and the development of secure, resilient autonomous systems.

--Building and deploying machine learning solutions in healthcare remains expensive and labor-intensive due to fragmented preprocessing workflows, model compatibility issues, and stringent data privacy constraints. In this work, we introduce an Agentic AI framework that automates the entire clinical data pipeline, from ingestion to inference, through a system of modular, task-specific agents. These agents are capable of handling both structured and unstructured data, enabling automatic feature selection, model selection, and preprocessing recommendation without manual intervention. We evaluate the system on publicly available datasets from geriatrics, palliative care, and colonoscopy imaging. For example, in the case of structured data (anxiety data) and unstructured data (colonoscopy polyps data), the pipeline begins with file-type detection by the "Ingestion Identifier Agent", followed by the "Data Anonymizer Agent" ensuring privacy compliance, where we first identify what type of data it is and then anonymize it. The "Feature Extraction Agent" then identifies features using an embedding-based approach for tabular data, which gives us all the column names, and a multistage MedGemma-based approach for image data, which gives us the modality and disease name. The "Preprocessing Recommender Agent" and "Preprocessing Implementor Agent" then apply tailored pre-processing based on data type and model requirements. Finally, the "Model Inference Agent" runs the selected model on the user uploaded data and generates interpretable outputs using tools like SHAP, LIME, and DETR attention maps. By automating these high-friction stages of the ML lifecycle, the proposed framework reduces the need for repeated expert intervention, offering a scalable and cost-efficient pathway for operationalizing AI in clinical environments. The integration of Artificial Intelligence (AI) into clinical workflows holds transformative potential for healthcare, enabling timely, data-driven decision-making across diagnosis and treatment planning [1].

Reinforcement learning (RL) has achieved remarkable success in fields like robotics and autonomous driving, but adversarial attacks designed to mislead RL systems remain challenging. Existing approaches often rely on modifying the environment or policy, limiting their practicality. This paper proposes an adversarial attack method in which existing agents in the environment guide the target policy to output suboptimal actions without altering the environment. We propose a reward iteration optimization framework that leverages large language models (LLMs) to generate adversarial rewards explicitly tailored to the vulnerabilities of the target agent, thereby enhancing the effectiveness of inducing the target agent toward suboptimal decision-making. Additionally, a critical state identification algorithm is designed to pinpoint the target agent's most vulnerable states, where suboptimal behavior from the victim leads to significant degradation in overall performance. Experimental results in diverse environments demonstrate the superiority of our method over existing approaches.