TechRAG: Evidence-Gated Multimodal Agentic RAG for Technical Literature Reasoning
Kanwar Bharat Singh · 2026
Read TechRAG: Evidence-Gated Multimodal Agentic RAG for Technical Literature Reasoning by Kanwar Bharat Singh (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-01613.txt.
Introduction
Singh presents TechRAG, an evidence-gated system for technical literature questions. The corpus holds several thousand papers on intelligent tires and vehicles. Conventional RAG runs a single pass over retrieved text. TechRAG runs an autonomous pipeline with an evidence gate instead. It classifies query intent and rewrites text and visual queries separately. Hybrid retrieval joins FAISS and BM25, then a cross-encoder reranks. A Neo4j knowledge graph expands evidence through chunk traversal. Visual evidence comes via late-interaction embeddings and MaxSim reranking.
A 100-point rubric scores whether evidence suffices for an answer. Low scores trigger drift-guarded retries of the retrieval. External databases get searched through vetting loops. Planner, Researcher, Writer, and Critic agents draft cited answers. Self-correcting revision closes the pipeline.
Catalog record
| CREATOR | Kanwar Bharat Singh |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 72 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.01613
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
TechRAG: Evidence-Gated Multimodal Agentic RAG for Technical Literature Reasoning Kanwar Bharat Singh Abstract This paper presents an agentic multimodal retrieval-augmented generation (RAG) framework for domain-specific literature reasoning, instantiated on a curated corpus of several thousand papers in intelligent tires, vehicle dynamics, vehicle control, sensing, estimation, and machine learning. Unlike conventional single-pass RAG systems, the proposed architecture uses an autonomous, evidence-gated pipeline that classifies query intent, generates separate text and visual query rewrites, performs hybrid text retrieval with FAISS and BM25 followed by cross-encoder reranking, expands evidence through graph-guided chunk traversal over a Neo4j knowledge graph, and retrieves visual document evidence using ColSmol late-interaction embeddings with MUVERA fixed-dimensional encoding, approximate nearest-neighbor search, and MaxSim reranking. The framework scores evidence sufficiency using a 100-point rubric with hybrid rule-based/LLM review, retries retrieval through drift-guarded reformulation, searches external academic databases through optimize–search–vet loops, merges and deduplicates multimodal evidence, verifies citation integrity, and generates cited answers through Planner, Researcher, Writer, and Critic agents with self-correcting revision. Key contributions include: (i) a scalable multimodal retrieval architecture combining text, graph, and visual evidence over 40,000 document pages; (ii) an interpretable evidence sufficiency and retry mechanism; (iii) a multi-agent generation pipeline with evidence mapping and critic-driven revision; (iv) a domain knowledge graph with LLM-based entity extraction, OpenAlex author validation, and intra-corpus citation resolution; and (v) a route-dependent external search architecture for targeted literature expansion. The result is a practical, evidence-gated, multimodal agentic RAG architecture for technical reasoning over specialized research corpora. 1 Introduction Recent advances in large language models (LLMs), enabled by the Transformer architecture 25 and scaled through models such as GPT-3 4 , have demonstrated significant potential to accelerate knowledge discovery, synthesis, and reasoning across a wide range of scientific and engineering domains. Techniques such as chain-of-thought prompting 26 ; 15 and tree-of-thought reasoning 29 have further enhanced the reasoning capabilities of these models. By enabling natural-language interaction with complex information, LLMs offer new opportunities to streamline research workflows and support technical decision-making. However, when applied to highly specialized technical fields, general-purpose LLMs exhibit fundamental limitations related to domain specificity, knowledge freshness, context scalability, and traceability. Over the past fifteen years, the author has been actively engaged in research and applied development across domains including intelligent tire systems, vehicle state estimation, vehicle dynamics, and vehicle control. This sustained involvement has necessitated continuous interaction with a broad and evolving body of specialized technical literature. As a natural outcome of this long-term research activity, the author has accumulated and curated a personal technical corpus comprising several thousand academic papers, yielding tens of thousands of indexed text chunks and approximately 40,000 document pages with visual content. The collection spans topics such as tire–road interaction modeling, intelligent tire sensing, vehicle and tire state estimation, sensor fusion, and advanced vehicle control algorithms, and has served as a foundational reference base supporting ongoing research and development efforts. While this curated collection represents a rich and authoritative knowledge base, its growing scale and interdisciplinary nature introduce significant practical challenges for effective utilization. Traditional keyword-based search tools and manual literature review processes become progressively less effective as corpus size and technical diversity increase. Relevant prior art, methodological insights, and cross-domain relationships are often obscured by inconsistent terminology, fragmented publication venues, and the cognitive burden associated with navigating large volumes of dense technical material. Critically, much of the valuable information in technical papers is conveyed through figures, charts, diagrams, and tables that are invisible to text-only retrieval systems. As a result, valuable knowledge may remain underutilized, despite being present within the available literature. The target users of such a system are engineers, researchers, and technical leaders working in research-heavy product development who need to connect technical literature and emerging research to generate evidence-backed insights, clarify the state of the art, drive design decisions, and support technical discussions. For these users, the ability to rapidly navigate a large body of literature with confidence in source attribution is essential. The motivation of this work is therefore to develop a scalable, domain-specific, multimodal technical knowledge engine capable of: • Systematically organizing and indexing large volumes of specialized literature across both text and visual content • Enabling semantic, meaning-based retrieval beyond conventional keyword search, including retrieval of relevant figures, charts, and diagrams directly from document pages • Rapidly surfacing relevant prior art, methods, and insights across adjacent technical domains • Autonomously evaluating evidence sufficiency and dynamically supplementing from external academic sources when internal evidence is insufficient • Maintaining structured knowledge representations through a graph database capturing relationships between papers, authors, methods, topics, and citations • Generating high-quality, cited answers through a multi-agent pipeline with structured planning, evidence mapping, and critic-driven self-correction • Supporting research, system design, and technical decision-making in intelligent tire and vehicle dynamics applications Standard RAG pipelines typically assume that retrieved chunks are sufficiently relevant once selected by the retriever. In technical literature reasoning, this assumption is fragile because relevant evidence may be incomplete, terminology may vary across subdomains, and retrieval confidence alone does not indicate whether the evidence can support an answer. Furthermore, standard text-only RAG pipelines miss the rich visual content—figures, charts, tables, and diagrams—that often conveys the most critical technical information. TechRAG addresses these gaps by making evidence sufficiency an explicit control signal, incorporating visual document retrieval alongside text retrieval, and employing multi-agent generation for structured, verifiable answer synthesis. To address these needs, this paper presents an agentic multimodal retrieval-augmented generation (RAG) framework tailored for domain-specific technical reasoning support. Drawing on advances in agentic LLM architectures that interleave reasoning with action 30 and learn from prior attempts through self-reflection 23 , multi-agent collaboration frameworks 27 ; 11 , and late-interaction visual document retrieval 9 , the proposed system goes beyond conventional single-pass RAG to employ a multi-step agentic pipeline. The pipeline autonomously classifies queries by intent, generates separate text and visual retrieval queries, performs hybrid text retrieval with cross-encoder reranking, expands retrieval via graph-guided chunk traversal, retrieves visually relevant document pages using ColSmol embeddings with MUVERA-based scalable search 7 , scores evidence sufficiency, performs agentic retry with drift-guarded reformulation, searches external academic databases through iterative loops, merges multimodal evidence, verifies citations, and generates answers through a Planner–Researcher–Writer–Critic multi-agent pipeline. By grounding model outputs in explicitly retrieved and vetted evidence from text, visual, graph, and external sources, the framework enables more reliable, transparent, and verifiable technical analysis. The central contribution of TechRAG is not a new retrieval model, foundation model, or graph algorithm, but a practical, evidence-gated, multimodal agentic architecture with interpretable control logic, domain-specific graph augmentation, and multi-agent generation that coordinates hybrid retrieval, visual document retrieval, graph traversal, external academic search, citation verification, and self-correcting multi-agent answer generation into a reproducible workflow for technical literature reasoning. This paper makes the following contributions: • A multimodal retrieval architecture that combines hybrid text retrieval (FAISS + BM25 + cross-encoder reranking), graph-guided chunk expansion via Neo4j topic and method node traversal, and visual document retrieval using ColSmol late-interaction embeddings with MUVERA fixed-dimensional encoding for scalable search over {\sim}40{,}000 document pages. • A multi-dimensional evidence sufficiency scoring framework (100-point rubric across five dimensions) with relevance damping, hybrid rule-based/LLM review, and agentic retry with drift-guarded reformulation, enabling the system to autonomously decide whether local evidence is sufficient or external