Computers & Industrial Engineering最新文献

{"title":"Design and implementation of a self-driving car using deep reinforcement learning: A comprehensive study","authors":"Rachid Djerbi,&nbsp;Anis Rouane,&nbsp;Zineb Taleb,&nbsp;Safia Saradouni","doi":"10.1016/j.cie.2025.111319","DOIUrl":"10.1016/j.cie.2025.111319","url":null,"abstract":"<div><div>This paper presents a groundbreaking and comprehensive study on the design, implementation, and evaluation of a self-driving car utilizing deep reinforcement learning, showcasing significant advancements in autonomous vehicle technology. Our robust framework integrates three innovative AI models for essential functionalities: road detection, traffic sign recognition, and obstacle avoidance. The system architecture, structured around a three layers “DDD” (Data, Detection, Decision) approach, involves meticulous data preprocessing for traffic signs and road data, followed by specialized Deep Learning models for each detection task, including a CNN for traffic signs, a CNN for road detection, and the pre-trained MobileNet-SSD for obstacle detection. A reinforcement learning agent in the Decision Layer processes these outputs for real-time control (steering, acceleration, braking) through a continuous learning process with environmental feedback. The research encompasses both extensive simulation in Unity, leveraging the ML-Agents toolkit for agent training across diverse environments, and crucial real-world deployment. Our reward/punishment system in the simulation environment, based on collisions with road markers and obstacles, refined the agent’s decision-making. The trained AI models were successfully exported and deployed onto a physical prototype, controlled by a Raspberry Pi and equipped with a camera and ultrasonic sensors. Real-world testing affirmed the robust performance of the physical model in detecting roads, recognizing traffic signs, and effectively avoiding obstacles. Quantitative results demonstrate compelling performance, including over 90% accuracy in obstacle detection and a 15% improvement in navigation efficiency compared to traditional algorithms under controlled simulation conditions. Model evaluation metrics show a 98% accuracy, 12% loss, and a prediction rate exceeding 77%. This study not only contributes a comprehensive framework for autonomous vehicle development but also highlights the transformative potential of deep reinforcement learning for creating intelligent and adaptable autonomous systems in both virtual and real-world scenarios, paving the way for safer and more efficient transportation technologies.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"207 ","pages":"Article 111319"},"PeriodicalIF":6.7,"publicationDate":"2025-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144513583","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Transient and steady-state analysis of multi-product serial lines with geometric machines","authors":"Xiaohan Wang ,&nbsp;Yaping Dai ,&nbsp;Bin Xin ,&nbsp;Zhiyang Jia ,&nbsp;Jiewu Leng","doi":"10.1016/j.cie.2025.111320","DOIUrl":"10.1016/j.cie.2025.111320","url":null,"abstract":"<div><div>The multi-product production line has become an important component of modern manufacturing. This study investigates a multi-product serial line with limited buffer capacity and machines obeying the geometric reliability model, analyzing both transient and steady-state production performance. We first provide an analytical analysis method for one- and two-machine lines. Then, for multi-machine lines, a computationally efficient approximation method is proposed based on an equivalent parameter calculation procedure that satisfies the production equivalence conditions. Numerical experiments validate the high accuracy of the proposed approximation method. Furthermore, the properties of the multi-product production line are investigated. Finally, a case study is presented to demonstrate the applicability of the proposed model and the analyzing approach.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"207 ","pages":"Article 111320"},"PeriodicalIF":6.7,"publicationDate":"2025-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144501547","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Interpretable accident prediction at highway-rail grade crossings: a deep learning approach","authors":"Xiang Yin,&nbsp;Jiangang Jin,&nbsp;Zhipeng Zhang","doi":"10.1016/j.cie.2025.111337","DOIUrl":"10.1016/j.cie.2025.111337","url":null,"abstract":"<div><div>Accidents at highway-rail grade crossings (HRGCs) pose significant risks to life and property, leading to substantial losses each year in the United States. Accurate and interpretable accident prediction provides a viable solution for improving the safety of HRGCs. Although encouraging processes have been achieved, existing studies either exhibit insufficient predictive performance or lack inherent interpretability, hindering efforts to further enhance the safety of HRGCs. To fill this gap, a well-designed deep learning model for accurate and interpretable accident prediction at HRGCs is proposed in this study. First, a word embedding approach is employed to generate vector representations of the category characteristics of HRGCs, effectively capturing the semantic information inherent in these characteristics. Second, the attention mechanism is used to separately aggregate the category characteristics and numerical characteristics, which can dynamically identify the key contributing characteristics of the accidents at HRGCs. The HRGCs data from Louisiana, Texas, and Washington were employed for a comparative analysis with the baseline model, demonstrating and validating the superiority and practicality of the proposed deep learning model. Finally, an interpretive analysis of the prediction process and prediction results of the proposed deep learning model is conducted. Eventually, this study explores the causative factors of accidents at HRGCs in a data-driven manner, providing valuable insights for further improving the safety performance of HRGCs.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"207 ","pages":"Article 111337"},"PeriodicalIF":6.7,"publicationDate":"2025-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489765","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An efficient method for solving large-scale open shop scheduling problem based on Horovod-GPU and improved graph attention network","authors":"Lanjun Wan ,&nbsp;Haoxin Zhao ,&nbsp;Xueyan Cui ,&nbsp;Long Fu ,&nbsp;Wei Ni ,&nbsp;Changyun Li","doi":"10.1016/j.cie.2025.111306","DOIUrl":"10.1016/j.cie.2025.111306","url":null,"abstract":"<div><div>The open shop scheduling problem (OSSP) involves complex processing constraints and a large number of job-machine combinations, which leads to an exponential increase in the solution space. For large-scale OSSP in real-world industrial productions, traditional methods struggle to provide satisfactory optimization results within a limited time. Therefore, an efficient method for solving large-scale OSSP through improved graph attention network based on link prediction (IGAT-LP) and Horovod-GPU is proposed. Firstly, an open shop scheduling (OSS) model based on IGAT-LP is designed to make full use of the feature information of operation nodes in OSSP. The model employs the graph attention network (GAT) structure to capture dependencies between tasks, learns global information through a multi-head attention mechanism, and predicts the optimal matching order between operations and machines. Secondly, a distributed parallelization method for the OSS model based on IGAT-LP is proposed. The distributed training capability of Horovod-GPU platform is fully utilized to expand the model training across multiple GPU nodes, significantly improving training efficiency. Finally, extensive experiments are conducted to analyze the effectiveness of the proposed method. The experimental results verify the superiority of the proposed method for solving large-scale OSSP instances. Moreover, the method significantly enhances the training performance of the OSS model based on IGAT-LP.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"207 ","pages":"Article 111306"},"PeriodicalIF":6.7,"publicationDate":"2025-06-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144471297","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Applications of autonomous driving technology in ride-hailing service platform: based on multi-party evolutionary game analysis","authors":"Deru Xie ,&nbsp;Huiqin Zhang ,&nbsp;Yuxiang Zhang ,&nbsp;Jiaman Yu","doi":"10.1016/j.cie.2025.111339","DOIUrl":"10.1016/j.cie.2025.111339","url":null,"abstract":"<div><div>The rapid development of autonomous driving technology is reshaping the business landscape of ride-hailing service platforms, and there is a gap in dynamic research on the multi-party collaborative promotion of autonomous driving technology. Therefore, the purpose of this study is to explore the strategic interactions among the government, autonomous driving technology providers, and ride-hailing service platforms in the promotion and application of autonomous driving technology; reveal the strategic dependencies of the three parties by constructing a multi-party dynamic evolution game model; and draw the following conclusions from the sensitivity analysis of the relevant parameters through numerical simulation: it is found that the initial willingness of the technology provider and the service provider is the determinant of the stability of the autonomous driving car technology promotion strategy; the market mechanism is the main driving force of technology promotion, and the government subsidy plays an auxiliary incentive role; the socio-economic benefits and the technology commission ratio are the key elements to achieve the stability strategy, and the optimization of the commission ratio can reduce the dependence of the two parties on the government subsidy; the government subsidy and the penalty mechanism can effectively accelerate the process of technology promotion, in which the government’s initial willingness to subsidize and the subsidy ratio have a significant positive effect on the willingness of ride-hailing service platforms to adopt the technology. The results of this study provide decision-making references for policymakers and stakeholders to adapt to the market changes and transformation brought by technology impacts and strategic suggestions for enterprises to grasp the window period of technology promotion.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"207 ","pages":"Article 111339"},"PeriodicalIF":6.7,"publicationDate":"2025-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144501546","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimizing integrated train rescheduling strategies for diverse disruption scenarios using reinforcement learning","authors":"Haodong Yin ,&nbsp;Lina Liu ,&nbsp;Ximing Chang ,&nbsp;Hao Fu ,&nbsp;Jianjun Wu","doi":"10.1016/j.cie.2025.111329","DOIUrl":"10.1016/j.cie.2025.111329","url":null,"abstract":"<div><div>Urban rail transit systems are crucial for efficient urban transportation, but unexpected disruptions pose significant challenges to maintaining optimal train schedules. Previous studies on train timetable rescheduling (TTR) have been limited by single disruption scenario and lack of comprehensive strategies, often relying on single or dual rescheduling tactics. This paper develops an innovative approach using reinforcement learning to optimize TTR under diverse disruption scenarios. Unlike previous studies, our approach incorporates a broader range of rescheduling strategies, including normal operations, train holding, short turning, reverse running, delayed departure from depots, and their combined strategies, enhancing the model’s flexibility and adaptability. To tackle the complexity of the proposed model, we further propose a novel reward mechanism with primary and secondary reward functions in reinforcement learning. The model’s effectiveness and adaptability are validated in various disruption scenarios using real-world cases on Beijing Metro Line 19. Experimental results demonstrate that, in a 30-minute disruption scenario, the integrated rescheduling strategy reduces total train delays by 46.6 % and computation time by 13.7 % compared to a single rescheduling strategy.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"207 ","pages":"Article 111329"},"PeriodicalIF":6.7,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489764","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessing economic and operational feasibility of a designed and lab demonstrated robotic platform for omnichannel logistics","authors":"Joyjit Bhowmick ,&nbsp;Sebastian Köhler ,&nbsp;Gideon Arndt ,&nbsp;Georg Fischer ,&nbsp;Manmit Padhy ,&nbsp;Kai Furmans ,&nbsp;Jennifer Pazour","doi":"10.1016/j.cie.2025.111304","DOIUrl":"10.1016/j.cie.2025.111304","url":null,"abstract":"<div><div>As omnichannel services, such as buy online pickup in store and home delivery, grow in popularity, many brick-and-mortar retailers have adopted a store fulfillment strategy, where the same inventory on the store shelves is used for both online and in-store customers. These omnichannel offerings shift the in-store logistics once done by shoppers to retailers. Thus, the focus of this work is to explore whether new material handling equipment has the potential to be deployed in a retail store environment to support omnichannel services. To do so, we designed and built a new picker-to-stock robotic platform to automate piece-level pick, sort, and place tasks in retail environments. Lab demonstrations of the robotic platform confirm the feasibility to robotically pick items from retail shelves and was able to achieve picking performance of 20 s per unit picked once in front of the shelf location. Then an agent-based simulation model is created to mimic a store’s logistical operations that integrates data from the robotic platform’s lab demonstrations and data from online and in-store customer demand. An iterative process determines the minimum amount of manual and robotic resources needed to operate the store that satisfies a given service level for online order fulfillment and replenishment tasks. Then to assess the economic viability of deploying such a robotic platform with the lab demonstrated values and with improved performance, these resource levels are then combined with operational metrics obtained from the simulation and various cost aspects via an economic analysis model. Computational experiments show that deploying the robotic platform for picking and restocking goods in a store environment is operationally and economically viable for retail grocery stores providing omnichannel services using a store fulfillment strategy.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"207 ","pages":"Article 111304"},"PeriodicalIF":6.7,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144519154","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimal order quantities of a multi-period inventory of compatible products","authors":"Doraid Dalalah","doi":"10.1016/j.cie.2025.111338","DOIUrl":"10.1016/j.cie.2025.111338","url":null,"abstract":"&lt;div&gt;&lt;div&gt;Effectively managing perishable inventory that consists of a heterogeneous assortment of mutually compatible products presents a complex and computationally challenging problem in operations research and supply chain fields. This particular problem arises when certain products can serve as substitutes for others, an operation commonly observed in maintenance parts, electronics, pharmaceuticals, batteries, and various other industries. Unlike single-period/single-product inventory models, where closed form solutions like the classical newsvendor are available in the literature, the compatibility between multiple products across multiple planning periods has remained largely unexplored in previous research.&lt;/div&gt;&lt;div&gt;Thus, the objective of our study is to find the optimal order quantities of different product types in an inventory system to minimize shortages and expiration, taking into account the challenging aspects of perishability and the existence of compatible substitutes. To rigorously address the complexities of this inventory system, a mixed-integer linear programming (MILP) framework is developed to encapsulate the intricate structural and operational characteristics of the problem. The proposed model is designed to accommodate both deterministic and stochastic scenarios, enabling the derivation of exact and approximate solutions through the integration of advanced optimization and simulation-based methodologies.&lt;/div&gt;&lt;div&gt;For large-scale instances, we recognize the computational challenges that arise when using standard solvers. Consequently, a metaheuristic algorithm is developed, which is specifically designed to reduce the computational time required to solve big instances. By tackling the interplay between perishability, compatibility, and multi periods, our study pushes the boundaries of existing research and presents innovative solution to real inventory where product substitution is allowed.&lt;/div&gt;&lt;div&gt;The findings showed that less shortages and expiration result in the case of compatible products. Compatibility also results in less order quantities. For a single inventory period, compatibility demonstrates higher effect on reducing the expected shortage and spoilage. For the case of full compatibility where any product can serve as a substitute for the others, closed form solution can be found for single period planning, while in multiple period planning, MILP optimization is required to address the problem. Finally, when compatibility is not present, it will lead to independent demand, thereby making the classical newsvendor problem applicable to this special case, but even in this case, optimization will be required for multi-period.&lt;/div&gt;&lt;div&gt;The proposed model exhibits broad applicability across a wide spectrum of industrial and medical applications, including but not limited to management of spare parts inventory, pharmaceutical supply chains, perishable food products such as dairy and ready-to-eat meals, the automo","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"207 ","pages":"Article 111338"},"PeriodicalIF":6.7,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144480834","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimizing matching radius for ride-hailing systems with dual-replay-buffer deep reinforcement learning","authors":"Jie Gao ,&nbsp;Rong Cheng ,&nbsp;Yaoxin Wu ,&nbsp;Honghao Zhao ,&nbsp;Weiming Mai ,&nbsp;Oded Cats","doi":"10.1016/j.cie.2025.111296","DOIUrl":"10.1016/j.cie.2025.111296","url":null,"abstract":"<div><div>The matching radius, defined as the maximum pick-up distance within which waiting riders and idle drivers can be matched, is a critical variable in ride-hailing systems. Optimizing the matching radius can significantly enhance system performance, but determining its optimal value is challenging due to the dynamic nature of ride-hailing environments. The matching radius should adapt to spatial and temporal variations, as well as to real-time fluctuations in supply and demand. To address this challenge, this paper proposes a dual-reply-buffer deep reinforcement learning method for dynamic matching radius optimization. By modeling the matching radius optimization problem as a Markov decision process, the method trains a policy network to adaptively adjust the matching radius in response to changing conditions in the ride-hailing system, thereby improving efficiency and service quality. We validate our method using real-world ride-hailing data from Austin, Texas. Experimental results show that the proposed method outperforms baseline approaches, achieving higher matching rates, shorter average pick-up distances, and better driver utilization across different scenarios.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"208 ","pages":"Article 111296"},"PeriodicalIF":6.7,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144572517","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Role of artificial intelligence in mitigating risk in multi-stage agricultural supply chain networks","authors":"Jiayi Zhu,&nbsp;Lei Yang,&nbsp;Lipan Feng","doi":"10.1016/j.cie.2025.111332","DOIUrl":"10.1016/j.cie.2025.111332","url":null,"abstract":"<div><div>More firms are adopting artificial intelligence (AI) to assist agricultural decision-making, but it is unclear whether and how AI can mitigate risk in agricultural supply chain. Driven by this question, we construct a multi-period and multi-layer agricultural supply chain network equilibrium model, where the theory of variational inequalities is utilized as the methodology. In this model, farmers make decisions at planting and harvest stages under yield uncertainty, and processors and retailers have to cope with potential disruptions through contingency measures. AI is introduced into supporting planting decision, and players’ interactions and risk cascade effects are incorporated. First, we find that AI enables farmers to shift from short-sightedness to a forward-looking perspective, which effectively stabilizes supply and price fluctuations, thereby mitigating risks. Second, AI makes the supply chain decisions for intertemporal products closer to those for non-intertemporal products, benefiting farmers, firms, and consumers. Third, AI can mitigate supply chain risks for non-intertemporal and low substitutable agricultural products more effectively than for other types of products. Finally, AI can also be a double-edged sword in specific scenarios, but the complementarity between AI and cold chain sharing enhances the efficiency of agricultural supply chains during unexpected disruptions. The findings provide valuable insights for the AI providers and managers to select the optimal agricultural products and develop incentive strategies.</div></div>","PeriodicalId":55220,"journal":{"name":"Computers & Industrial Engineering","volume":"207 ","pages":"Article 111332"},"PeriodicalIF":6.7,"publicationDate":"2025-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489763","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}