arXiv subjects

Manan Mehta

Publications and source records attributed to Manan Mehta.

11 recordsLinked to original sources

arXiv2026

NavTrust: Benchmarking Trustworthiness for Embodied Navigation

There are two major categories of embodied navigation: Vision-Language Navigation (VLN), where agents navigate by following natural language instructions; and Object-Goal Navigation (OGN), where agents navigate to a specified target object. However, existing work primarily evaluates model performance under nominal conditions, overlooking the potential corruptions that arise in real-world settings. To address this gap, we present NavTrust, a unified benchmark that systematically corrupts input modalities, including RGB, depth, and instructions, in realistic scenarios and evaluates their impact on navigation performance. To our best knowledge, NavTrust is the first benchmark that exposes embodied navigation agents to diverse RGB-Depth corruptions and instruction variations in a unified framework. Our extensive evaluation of seven state-of-the-art approaches reveals substantial performance degradation under realistic corruptions, which highlights critical robustness gaps and provides a roadmap toward more trustworthy embodied navigation systems. Furthermore, we systematically evaluate four distinct mitigation strategies to enhance robustness against RGB-Depth and instruction corruptions. Our base models include Uni-NaVid and ETPNav. We deployed them on a real mobile robot and observed improved robustness to corruptions. The project website is: https://navtrust.github.io/.

cs.RO↗

arXiv2026

A Unified Hierarchical Multi-Task Multi-Fidelity Framework for Data-Efficient Surrogate Modeling in Manufacturing

Surrogate modeling is an essential data-driven technique for quantifying relationships between input variables and system responses in manufacturing and engineering systems. Two major challenges limit its effectiveness: (1) large data requirements for learning complex nonlinear relationships, and (2) heterogeneous data collected from sources with varying fidelity levels. Multi-task learning (MTL) addresses the first challenge by enabling information sharing across related processes, while multi-fidelity modeling addresses the second by accounting for fidelity-dependent uncertainty. However, existing approaches typically address these challenges separately, and no unified framework simultaneously leverages inter-task similarity and fidelity-dependent data characteristics. This paper develops a novel hierarchical multi-task multi-fidelity (H-MT-MF) framework for Gaussian process-based surrogate modeling. The proposed framework decomposes each task's response into a task-specific global trend and a residual local variability component that is jointly learned across tasks using a hierarchical Bayesian formulation. The framework accommodates an arbitrary number of tasks, design points, and fidelity levels while providing predictive uncertainty quantification. We demonstrate the effectiveness of the proposed method using a 1D synthetic example and a real-world engine surface shape prediction case study. Compared to (1) a state-of-the-art MTL model that does not account for fidelity information and (2) a stochastic kriging model that learns tasks independently, the proposed approach improves prediction accuracy by up to 19% and 23%, respectively. The H-MT-MF framework provides a general and extensible solution for surrogate modeling in manufacturing systems characterized by heterogeneous data sources.

cs.LG↗

arXiv2025

RLHF: A comprehensive Survey for Cultural, Multimodal and Low Latency Alignment Methods

Reinforcement Learning from Human Feedback (RLHF) is the standard for aligning Large Language Models (LLMs), yet recent progress has moved beyond canonical text-based methods. This survey synthesizes the new frontier of alignment research by addressing critical gaps in multi-modal alignment, cultural fairness, and low-latency optimization. To systematically explore these domains, we first review foundational algo- rithms, including PPO, DPO, and GRPO, before presenting a detailed analysis of the latest innovations. By providing a comparative synthesis of these techniques and outlining open challenges, this work serves as an essential roadmap for researchers building more robust, efficient, and equitable AI systems.

cs.LG↗

arXiv2025

Small Language Models for Agentic Systems: A Survey of Architectures, Capabilities, and Deployment Trade offs

Small language models (SLMs; 1-12B params, sometimes up to 20B) are sufficient and often superior for agentic workloads where the objective is schema- and API-constrained accuracy rather than open-ended generation. We synthesize recent evidence across open and proprietary SLMs (Phi-4-Mini, Qwen-2.5-7B, Gemma-2-9B, Llama-3.2-1B/3B, Ministral-3B/8B, Apple on-device 3B, DeepSeek-R1-Distill) and connect it to modern evaluations (BFCL v3/v4, StableToolBench) and serving stacks (vLLM, SGLang, TensorRT-LLM) paired with guided decoding libraries (XGrammar, Outlines). We formalize SLM-default, LLM-fallback systems with uncertainty-aware routing and verifier cascades, and propose engineering metrics that reflect real production goals: cost per successful task (CPS), schema validity rate, executable call rate, p50/p95 latency, and energy per request. Guided decoding, strict JSON Schema outputs, and validator-first tool execution close much of the capability gap with larger models and often let SLMs match or surpass LLMs on tool use, function calling, and RAG at 10x-100x lower token cost with materially better latency and energy. We provide design patterns for agent stacks that prioritize SLMs: schema-first prompting, type-safe function registries, confidence scoring with verifier rollups, and lightweight adaptation via LoRA/QLoRA. We also delineate limits where fallback remains valuable (open-domain reasoning and some long-horizon planning). The result is a practical blueprint for building fast, inexpensive, and reliable agents that default to SLMs while preserving headroom with targeted LLM assistance. Keywords: small language models, agents, function calling, structured outputs, JSON Schema, guided decoding, LoRA/QLoRA, routing, energy efficiency, edge inference

cs.AI↗

arXiv2025

Adaptive and Explainable AI Agents for Anomaly Detection in Critical IoT Infrastructure using LLM-Enhanced Contextual Reasoning

Ensuring that critical IoT systems function safely and smoothly depends a lot on finding anomalies quickly. As more complex systems, like smart healthcare, energy grids and industrial automation, appear, it is easier to see the shortcomings of older methods of detection. Monitoring failures usually happen in dynamic, high dimensional situations, especially when data is incomplete, messy or always evolving. Such limits point out the requirement for adaptive, intelligent systems that always improve and think. LLMs are now capable of significantly changing how context is understood and semantic inference is done across all types of data. This proposal suggests using an LLM supported contextual reasoning method along with XAI agents to improve how anomalies are found in significant IoT environments. To discover hidden patterns and notice inconsistencies in data streams, it uses attention methods, avoids dealing with details from every time step and uses memory buffers with meaning. Because no code AI stresses transparency and interpretability, people can check and accept the AI's decisions, helping ensure AI follows company policies. The two architectures are put together in a test that compares the results of the traditional model with those of the suggested LLM enhanced model. Important measures to check are the accuracy of detection, how much inaccurate information is included in the results, how clearly the findings can be read and how fast the system responds under different test situations. The metaheuristic is tested in simulations of real world smart grid and healthcare contexts to check its adaptability and reliability. From the study, we see that the new approach performs much better than most existing models in both accuracy and interpretation, so it could be a good fit for future anomaly detection tasks in IoT

cs.AI↗

arXiv2014

A real-time software simulator for scanning force microscopy

We describe software that simulates the hardware of a scanning force microscope. The essential feature of the software is its real-time response, which is critical for mimicking the behavior of real scanning probe hardware. The simulator runs on an open-source real time Linux kernel, and can be used to test scanning probe microscope control software as well as theoretical models of different types of scanning probe microscopes. We describe the implementation of a tuning-fork based atomic force microscope and a dc electrostatic force microscope, and present representative images obtained from these models.

cond-mat.mes-hall↗

arXiv2013

A Hybrid Analog/Digital Phase-Locked Loop for Frequency Mode Non-contact Scanning Probe Microscopy

Non-contact scanning probe microscopy (SPM) has developed into a powerful technique to image many different properties of samples. The conventional method involves monitoring the amplitude, phase or frequency of a cantilever oscillating at or near its resonant frequency as it is scanned across the surface of a sample. For high Q factor cantilevers, monitoring the resonant frequency is the preferred method in order to obtain reasonable scan times. This can be done by using a phase-locked-loop (PLL). PLLs can be obtained as commercial integrated circuits, but these do not have the frequency resolution required for SPM. To increase the resolution, all-digital PLLs requiring sophisticated digital signal processors or field programmable gate arrays have also been implemented. We describe here a hybrid analog/digital PLL where most of the components are implemented using discrete analog integrated circuits, but the frequency resolution is provided by a direct digital synthesis chip controlled by a simple PIC microcontroller. The PLL has excellent frequency resolution and noise, and can be controlled and read by a computer via a USB connection.

cond-mat.mes-hall↗

arXiv2012

RTSPM: Real-time Linux control software for scanning probe microscopy

Real time computer control is an essential feature of scanning probe microscopes, which have become essential tools for the characterization and investigation of nanometer scale samples. Most commercial (and some open-source) scanning probe data acquisition software uses digital signal processors (DSPs) to handle the real time data processing and control, which adds to the expense and complexity of the control software. We describe here scan control software that uses a single computer and a data acquisition card to acquire scan data. The computer runs an open-source real time Linux kernel, which permits fast acquisition and control while maintaining a responsive graphical user interface. Images from a simulated tuning-fork based microscope as well as a standard topographical sample are also presented, showing some of the capabilities of the software.

cond-mat.mes-hall↗

arXiv2012

Evidence for charge-vortex duality at the LaAlO$_3$/SrTiO$_3$ interface

The concept of duality has proved extremely powerful in extending our understanding in many areas of physics (1, 2). Charge-vortex duality has been proposed (3, 4) as a model to understand the superconductor to insulator transition (SIT) in disordered thin films (5, 6) and Josephson junction arrays (7, 8, 9). In this model, on the superconducting side, one has delocalized Cooper pairs but localized vortices; while on the insulating side, one has localized Cooper pairs but mobile vortices. Here we show a new experimental manifestation of this duality in the electron gas that forms at the interface between LaAlO$_3$ (LAO) and SrTiO$_3$ (STO) (10, 11, 12, 13, 14). The effect is due to the motion of vortices generated by the magnetization dynamics of the ferromagnet that also forms at the same interface (14, 15, 16), which results in an increase in resistance on the superconducting side of the transition, but an increase in conductance on the insulating side.

cond-mat.supr-con↗

arXiv2010

Conductance asymmetry in point-contacts on epitaxial thin films of Ba(Fe$_{0.92}$Co$_{0.08}$)$_2$As$_2$

Point-contact spectroscopy is a powerful tool for probing superconductors. One of the most common observations in the point-contact spectra on the recently discovered ferropnictide superconductors is a large conductance asymmetry with respect to voltage across the point-contact. In this paper we show that the antisymmetric part of the point-contact spectrum between a silver tip and an epitaxial thin film of Ba(Fe$_{0.92}$Co$_{0.08}$)$_2$As$_2$ shows certain unique features. These features have an interesting evolution with increasing temperature up to a temperature that is 30% larger than the critical temperature $T_c$ of the superconductor. We argue that this evolution can be associated with the rich normal state properties of these materials.

cond-mat.supr-con↗

arXiv2010

Phase-incoherent superconducting pairs in the normal state of Ba(Fe$_{1-x}$Co$_x$)$_2$As$_2$

The normal state properties of the recently discovered ferropnictide superconductors might hold the key to understanding their exotic superconductivity. Using point-contact spectroscopy we show that Andreev reflection between an epitaxial thin film of Ba(Fe$_{0.92}$Co$_{0.08}$)$_2$As$_2$ and a silver tip can be seen in the normal state of the film up to temperature $T\sim1.3T_c$, where $T_c$ is the critical temperature of the superconductor. Andreev reflection far above $T_c$ can be understood only when superconducting pairs arising from strong fluctuation of the phase of the complex superconducting order parameter exist in the normal state. Our results provide spectroscopic evidence of phase-incoherent superconducting pairs in the normal state of the ferropnictide superconductors.

cond-mat.supr-con↗