An Agentic AI Framework with Large Language Models and Chain-of-Thought for UAV-Assisted Logistics Scheduling with Mobile Edge Computing

Hanwen Zhang, Dusit Niyato, Wei Zhang · 2026

Read An Agentic AI Framework with Large Language Models and Chain-of-Thought for UAV-Assisted Logistics Scheduling with Mobile Edge Computing by Hanwen Zhang, Dusit Niyato, Wei Zhang (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-2605-13221.txt.

Introduction

Zhang and colleagues design an agent framework that plans drone logistics together with mobile edge computing. Drones collect finished goods from factory stations while also carrying computing for sensor tasks, so routes and computing schedules interact. A language model agent turns user requests into a clear maths formulation, and a two layer learning method then plans routes above and task slots below. Simulations reach full product collection in nearly all late episodes with every task deadline met.

Catalog record

CREATORHanwen Zhang, Dusit Niyato, Wei Zhang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE78000
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.13221

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

An Agentic AI Framework with Large Language Models and Chain-of-Thought for UAV-Assisted Logistics Scheduling with Mobile Edge Computing Thanks: This work was supported in part by the Agency for Science, Technology and Research (A*STAR) under its Manufacturing, Trade & Connectivity (MTC) Individual Research Grants (IRG) under Grant M23M6c0113; in part by MTC Programmatic under Grant M23L9b0052; in part by the Seatrium New Energy Laboratory, Singapore Ministry of Education (MOE) Tier 1, under Grant RT5/23 and Grant RG24/24; in part by the Nanyang Technological University (NTU) Center for Computational Technologies in Finance (NTU-CCTF); and in part by the Research Innovation and Enterprise (RIE) 2025 Industry Alignment Fund-Industry Collaboration Projects (IAF-ICP), administered by A*STAR, under Grant I2301E0026. ( Corresponding author: Wei Zhang.) Thanks: Hanwen Zhang is with both the College of Computing and Data Science, Nanyang Technological University, Singapore 639798, and the Information and Communications Technology Cluster, Singapore Institute of Technology, Singapore 828608 (e-mail: [email protected] and [email protected]) Thanks: Dusit Niyato is with the College of Computing and Data Science, Nanyang Technological University, Singapore 639798 (e-mail: [email protected]). Thanks: Wei Zhang, Xin Lou, and Malcolm Yoke Hean Low are with the Information and Communications Technology Cluster, Singapore Institute of Technology, Singapore 828608 (e-mail: [email protected], [email protected], and [email protected]). Hanwen Zhang, Dusit Niyato, Wei Zhang Abstract In cloud manufacturing, unmanned aerial vehicles (UAVs) can support both product collection and mobile edge computing (MEC). This joint operation forms a hybrid scheduling problem, where physical logistics decisions are coupled with computational task scheduling. In this paper, UAVs collect finished products from manufacturing stations and transport them back to a central depot. Meanwhile, computational tasks generated by industrial sensor devices at these stations are processed locally, at UAVs, or offloaded via UAVs to the cloud. This coupling makes the problem challenging. A UAV can provide MEC services only during its service window at a station, so routing decisions directly determine when UAV-assisted offloading is available. Routing decisions also affect the UAV energy budget and the availability of onboard computing and communication resources for computational task execution under task deadline constraints. To address this, we propose an agentic-AI-assisted optimization framework with two components. First, we develop an agentic AI that combines large language models, retrieval-augmented generation, and chain-of-thought reasoning to translate user input into an interpretable mathematical formulation for the hybrid scheduling problem. Second, we design a hierarchical deep reinforcement learning approach based on proximal policy optimization (PPO), where the upper layer learns UAV routing and the lower layer optimizes per-slot task execution and resource allocation. Simulation results show that the proposed framework yields more consistent formulations, while the hierarchical PPO achieves full product collection in 99.6% of the last 500 episodes and maintains a 100% deadline satisfaction rate, with more stable performance than the advantage actor-critic approach. \DeclareAcroListStyle mydesclist list = description , I Introduction As large language models (LLMs) are progressing rapidly, the adoption of LLMs in manufacturing applications is continuously increasing. This is especially true for scheduling problems that require semantic understanding and multi-step decision support. For example, the authors in [ 52 ] developed an LLM-based multi-agent scheduling chain for the flexible job shop scheduling problem. In this framework, an improved decision-making process is used to analyze scheduling indicators and select suitable algorithms. The overall framework also supports dynamic scheduling and real-time rescheduling. More recently, [ 18 ] proposed an LLM-empowered dynamic scheduling framework for intelligent hybrid flow shops. In this framework, structured prompt engineering is used to enrich state representation and guide action selection within a multi-agent deep reinforcement learning (DRL) framework. LLMs have also been utilized to improve the performance of solving combinatorial optimization problems [ 20 ] . LLMs have been applied to tackle the job shop scheduling problem [ 1 ] , and integrated into a multi-agent manufacturing system to handle dynamic changes in production tasks [ 56 ] . Moreover, cloud manufacturing (CMfg), a key paradigm of smart manufacturing, enables service-oriented sharing and efficient allocation of geographically dispersed resources, where coordinated decision-making across cyber-physical components is essential to achieve its value [ 5 ] . With the continued advancement of Internet of Things (IoT) technologies, CMfg increasingly relies on cloud-edge-device collaboration, in which computational tasks (e.g., digital twins, real-time analytics, and predictive maintenance) play a critical role in supporting time-sensitive industrial operations [ 50 ] . In a typical CMfg setting, physical logistics operations can be executed by automated agents (e.g., unmanned aerial vehicles (UAVs) [ 33 ] or automated guided vehicles [ 32 ] ), while computational workloads are processed at the device, edge, or cloud layers [ 44 ] . These computational workloads are typically generated by industrial sensing and monitoring applications, such as image-recognition-based quality inspection, anomaly detection, real-time analytics, and predictive maintenance. In this paper, we study an integrated scheduling problem that jointly optimizes UAV-assisted product collection and cross-layer computational task offloading from devices to UAVs and the cloud. This joint scheduling problem is a challenging combinatorial optimization task due to the tight coupling between UAV routing/service availability and task execution decisions under stringent resource and deadline constraints, as discussed below. Challenge I : The mathematical modeling of combinatorial optimization problems in CMfg can be significantly more complex than that of traditional single-domain scheduling problems. Similar coupling complexity has also been observed in UAV-assisted MEC systems, where computational task offloading is jointly optimized with UAV trajectory, path-planning, or task-scheduling decisions [ 21 ] . It requires capturing how UAV routing induces time-varying service windows. These service windows, in turn, affect per-slot task offloading and compute/communication allocation under energy budgets, capacity limits, and task deadlines. At the same time, the model must balance conflicting goals, such as logistics efficiency and sufficient resource allocation for meeting task deadlines. These complexities lead to a complicated modeling process. As a result, the problem becomes challenging for engineers and researchers to model, formulate, and solve such complex problems. This issue is due to the fact that specific operational dynamics of joint logistics-computation scheduling CMfg have to be taken into account. Challenge II : Compared with traditional scheduling problems, hybrid task scheduling in CMfg introduces significantly higher solution complexity. The reason is that logistics- and computation-related decisions must be optimized jointly rather than handled in isolation. Moreover, as user demands and system conditions evolve over time, the resulting optimization problem may also change. Therefore, the key challenge is to solve these changing optimization problems efficiently. In particular, the scheduler must jointly determine UAV routing, offloading-destination selection, and compute/communication allocation under strict latency, resource, and communication constraints. Therefore, designing DRL frameworks that can generalize across different hybrid task scenarios while mitigating sparse feedback and complex action representation is crucial for enhancing the efficiency and adaptability of intelligent CMfg systems. To formulate complex combinatorial problems in CMfg, we present an agentic AI framework to address Challenge I . The agent facilitates dialogue-driven mathematical model generation and refinement by combining LLM reasoning with retrieval augmented generation (RAG) and CoT reasoning. In particular, an LLM converts natural language descriptions into structured modeling logic. Then, RAG extracts pertinent formulations, constraints, and objectives from curated literature and technical documentation. CoT further makes explicit the logic underlying variable definitions, constraint design, and objective construction. Hence, CoT can enhance the accuracy

Read the complete text