（原标题：红绿灯背后的AI大模型：从被动响应到全局预判） 南方财经见习记者郭莎 广州报道 12月12日至14日，2025年粤港澳大湾区人工智能与机器人产业大会暨广东省人工智能与机器人技能大赛 在广州举行。其中，12日举行的"AIR＋交通"论坛以"智驭交通·未来之路"为主题，各界嘉宾聚焦AI技术 如何驱动智慧交通系统升级、面临哪些商业化挑战，以及如何构建支撑长远发展的产业生态等议题，展 开了深度对话与思想碰撞。 "AI大模型正让交通信号控制从'被动响应'转向'主动预测'。""人形机器人在交通领域具备结构化场景、 环境稳定、市场广阔等独特优势，有望率先实现规模应用。""我们必须在人才培养上推动学科深度交 叉，让学生既懂交通专业，又掌握AI技术。"圆桌讨论环节，来自学界与业界的多位专家金句频出。 佳都科技智慧城市交通业务群常务副总裁许健以生动的"下棋"比喻阐释了AI带来的变革。他分析，传统 的信号控制如同"对手下一步，我下一步"的被动响应，而如今基于海量交通数据训练的大模型，则让系 统具备了"学习棋谱、预判未来"的能力。 据介绍，广州6个城区的主要路口都使用了AI大模型来生成红绿灯的配置和控制方案，通行效率大幅提 ...

另一项成熟应用是城市交通信号优化。 王凯指出，由大模型驱动的信控智能体，已能自主生成并下发 红绿灯配时方案。该应用从广州珠江新城十几个路口的试点，已扩展至六个主城区，实现区域级24小时 不间断优化，有效提升了通行效率，并已推广至重庆、西安等城市。 当强对流天气即将来袭，AI已提前"读"懂天气与舆情，为机场调度好应急资源；当你驾车经过路口，红 绿灯配时可能正由云端"智能体"实时优化。12月12日，2025年粤港澳大湾区人工智能与机器人产业大会 （XAIR）——"AIR+交通"专场活动暨人工智能赋能新型工业化供需对接深度行（广州站）活动上，专 家指出，交通智能化已不止于"感知反应"，更在于"预测决策"。场景丰富的广东，正成为实践这一前沿 理念的核心阵地。 省委统战部副部长、省工商联党组书记陈志清在致辞中表示，"广东智造"的交通解决方案，正从湾区走 向全国、走向世界，佳都科技等优秀企业在智慧轨道、智慧交通等领域快速成长。他表示，期待广大民 营企业勇担重任、走在前列，发挥示范引领作用。省工商联将继续发挥桥梁纽带功能，积极搭建服务平 台，支持民营企业加强科技创新，推动智能交通产业高质量发展。 "智慧交通的未来必将迈向' ...

Core Insights - The event focused on the deep integration and innovative application of AI and robotics in the transportation sector, highlighting the importance of collaboration among government, industry, academia, and research [2][4]. Group 1: Event Overview - The 2025 Guangdong-Hong Kong-Macao Greater Bay Area Artificial Intelligence and Robotics Industry Conference featured a special session on "AIR + Transportation," showcasing advanced technologies like the "Traffic Jia Hong" operating system and all-station cruising robots [4]. - The event aimed to facilitate the transformation of technological achievements into productive forces through product displays, project roadshows, and specialized discussions [4][5]. Group 2: Industry Development - The Vice Minister of the Guangdong Provincial United Front Work Department emphasized that Guangdong's transportation solutions are expanding from the Greater Bay Area to the entire country and globally, with several companies rapidly growing in smart rail and smart transportation sectors [4]. - The event included a "Supply and Demand Matching Symposium," where representatives discussed urban traffic governance and urban rail transit, leading to initial cooperation intentions [7]. Group 3: Expert Insights - A keynote report by a Singapore Academy of Engineering Fellow analyzed the latest international research progress and future trends in sensor technology and multimodal perception technology in areas such as autonomous driving and traffic management [7]. - The President of the Guangdong Intelligent Transportation Association discussed the transition of intelligent transportation from "perception and reaction" to "awareness intelligence," outlining key pathways for this transformation [7].

Group 1 - The 2025 Guangdong-Hong Kong-Macao Greater Bay Area Artificial Intelligence and Robotics Industry Conference (XAIR) focuses on the integration of AI in transportation, emphasizing the need for an open, collaborative, and sustainable ecosystem [1] - The project led by the China Industrial Internet Research Institute aims to create a comprehensive service platform that covers technology research and development, scenario matching, and industrial collaboration, accelerating the integration of AI solutions in key sectors like transportation and manufacturing [1] - The event aims to establish efficient connections between government, industry, academia, research, and investment, facilitating precise matching of demand and technology, and promoting high-quality development in transportation through AI [1] Group 2 - The Chief AI Scientist of Jiadu Technology presented a dual-driven approach using the "Zhixing" industry model and "Traffic Jiahong" operating system to build a smart transportation empowerment system, focusing on real-time network status perception and operational efficiency [2] - The President of the Guangdong Intelligent Transportation Association highlighted the transition of intelligent transportation from "perception and reaction" to "awareness intelligence," which includes predictive and decision-making capabilities, outlining key pathways for this transformation [2] - The event showcased practical applications of AI technologies, such as the "Road Xiaoling Tong" intelligent monitoring and early warning system for highways, demonstrating real-time network status perception and enhanced management efficiency [2] Group 3 - The "AIR+ Transportation" event effectively promoted the deep integration of AI and robotics technology with the transportation industry, contributing to the formation of new productive forces in the transportation sector and the establishment of a modern comprehensive transportation system for the future [3]

Core Viewpoint - The article discusses the strategic transformation and technological innovations of Jiadu Technology, emphasizing its role in the "AI + Urban" sector and the development of the "Traffic Jiahong" operating system based on the HarmonyOS, which aims to redefine future station operations and enhance the efficiency of urban transportation systems [1][2]. Group 1: Company Strategy and Development - Jiadu Technology has consistently aligned its strategic transformations with national policies, technological trends, and its core capabilities, which has been crucial for its sustained growth [2][3]. - The company has focused on digital economy, new infrastructure, and the national strategy of building a strong transportation system, particularly following the government's recent initiatives on "AI +" [2][3]. - Jiadu Technology's self-developed "Zhixing Transportation Model" is noted as the world's first large model in the transportation sector, with plans to release the "Traffic Jiahong" operating system in 2025 [2][3]. Group 2: Technological Innovations - The company is addressing industry challenges through high R&D investment and strategic mergers and acquisitions, creating a competitive edge by integrating technology and capital [3][4]. - Jiadu Technology is developing a digital foundation for transportation, focusing on creating a domestic operating system and basic software to enhance industry capabilities [4][5]. - The introduction of the "Zhixing Transportation Model 3.0" utilizes FP8 mixed precision training technology to improve training efficiency without sacrificing accuracy [5]. Group 3: Application and Impact - Jiadu Technology's innovations have been successfully implemented in various benchmark projects, such as the Guangzhou Metro and Chengdu's smart city traffic management, achieving significant efficiency improvements [7][9]. - The company is expanding its services beyond government sectors to include digital upgrades for businesses in finance, healthcare, and retail, with over 10,000 successful delivery cases [9][10]. - Jiadu Technology plans to initiate its global expansion in 2024, targeting regions with high infrastructure demand, such as countries along the Belt and Road Initiative and RCEP [10]. Group 4: Future Outlook - The company anticipates a shift in the "AI + Urban" landscape towards a more integrated model, leveraging key technologies like urban large models and digital twins for comprehensive urban management [10]. - Jiadu Technology aims to enhance its R&D investments in frontier technologies and deepen the integration of digital technology with the real economy to create measurable customer and social value [10].

Core Insights - Company is seizing opportunities in the deep integration of artificial intelligence and the transportation industry, enhancing its "JiaDu ZhiXing Transportation Model" technology advantages and launching the first domestic operating system in the transportation sector, "JiaTian JiaHong" [1][2] - In the first half of 2025, the company achieved operating revenue of 4.919 billion, a year-on-year increase of 65.22%, and a net profit attributable to shareholders of 135 million, marking a return to profitability [1] - The company focuses on full-stack AI technology products and solutions, leveraging core technologies such as computer vision and data science to create a digital foundation for urban transportation [1] Product Development - The "JiaTian JiaHong" operating system is based on OpenHarmony and openEuler, addressing core needs in the transportation industry such as high reliability and real-time data processing [2] - The system aims to resolve industry pain points like fragmented interconnectivity protocols and data silos, covering urban and rail transportation scenarios [2] - The company plans to increase investment in "JiaTian JiaHong" technology development and collaborate with transportation management departments and technology suppliers for tailored solutions [2] Policy Impact - The recent "Artificial Intelligence+" policy provides a framework for the company to deepen AI application scenarios and promote smart technology in intelligent rail transportation and urban traffic [3] - High-quality scene data, such as real-time traffic flow and rail operation logs, are identified as core resources for model training and industry implementation [3] - The company aims to leverage the policy to enhance its leading position in the smart transportation sector [3] Strategic Partnerships - As a key player in Guangdong's strategic industrial cluster, the company drives innovation through a four-chain model involving industry, innovation, finance, and talent [4] - The company has invested in several AI technology firms and established research collaborations with prestigious institutions like the Chinese Academy of Sciences and Tsinghua University [4] - The goal is to create a high-tech industrial cluster with strong demonstration capabilities, forming a "JiaDu+" AI industry community [4]

Core Insights - The transportation industry serves as a clear window into the progress of urban intelligence, with China's urban passenger volume reaching 1,067.97 billion trips by the end of 2024 [1] - The demand for smarter and more efficient solutions is driven by persistent issues such as congestion and safety hazards, prompting urban managers to seek advanced technologies [1] - JiaDu Technology Group Co., Ltd. has established itself as a leader in AI-driven transportation solutions, focusing on "smart rail transit + smart urban transportation" [1][3] Industry Overview - The Chinese urban transportation system is entering a new phase of digital intelligence upgrade, with "smart rail transit and smart urban transportation" at its core [2] - National policies, such as the "14th Five-Year Plan" and the "Guidelines for the Development of Smart Urban Rail Transit," support the acceleration of smart rail transit system construction [2] - The market for smart urban transportation systems has grown from 62.8 billion yuan in 2020 to an estimated 83.6 billion yuan in 2024, with projections to reach 147.1 billion yuan by 2029 [2] Company Positioning - JiaDu Technology has over 30 years of experience in the smart transportation sector, leveraging proprietary AI models and industrial control technologies [3] - The company has developed a comprehensive product ecosystem covering the entire lifecycle of rail transit systems, including core products like AFC, PSD, ISCS, and communication systems [4] - JiaDu Technology's solutions have been implemented in 45 cities, covering over 110 metro lines and more than 2,200 stations, accounting for one-third of the total metro mileage in China [5] Technological Advancements - The company has created the world's first AI smart station in collaboration with Guangzhou Metro, enhancing operational efficiency and passenger experience [4] - JiaDu Technology's IDPS urban traffic brain solution integrates AI, digital twins, and big data, providing comprehensive governance capabilities across major urban areas [5] - The company's revenue from smart urban transportation solutions reached 2.321 billion yuan by the end of 2024, reflecting a year-on-year growth of 16.75% [6] Competitive Edge - JiaDu Technology's dual-core technology system includes the "Traffic JiaHong" operating system and the "JiaDu ZhiXing" large model, addressing industry challenges such as data silos and operational inefficiencies [7] - The company has a robust R&D team exceeding 2,700 personnel and has filed over 2,000 patents, establishing a strong technological barrier [7] - JiaDu Technology has been recognized in the "2024 Hurun China AI Top 50" and won a silver medal at the 50th Geneva International Exhibition of Inventions, highlighting its innovative capabilities [8] Future Outlook - The company aims to solidify its leading position in the smart transportation industry through its differentiated strategic positioning and advanced technological strength, especially following its anticipated listing on the Hong Kong Stock Exchange [9]

Core Viewpoint - The next three years are critical for Jiadu Technology, focusing on continuous investment in the intelligent transportation sector and strengthening its technological foundation [2] Group 1: Technological Development - Jiadu Technology aims to build a full-stack competitive advantage in the intelligent transportation field through continuous technological breakthroughs and self-reliance [2] - The company has adopted an "All in AI" strategy, focusing on AI innovations to empower the industry and commercialize technologies like AI large models [3] - Jiadu Technology has launched the first domestic operating system in the transportation industry, "Jiaotong Jiahong," to establish a self-controlled industry operating system [4] - The company has developed a dual-core technology system with "Jiaotong Jiahong" as the digital foundation and "Jiadu Zhixing Transportation Large Model" as the intelligent application brain [4] - Jiadu Technology has completed the adaptation verification of advanced models like DeepSeek-V3 and R1, deploying them in over 10 cities' intelligent transportation projects [4] - The company maintains a leading position in smart technology through significant R&D investments, particularly in rail transportation and operational energy-saving applications [4][5] Group 2: Financial Performance - In 2024, Jiadu Technology's R&D investment is projected to be 243 million yuan, with a total of 89 patent applications and 41 patents granted [5] - In the first half of 2023, Jiadu Technology achieved operating revenue of 4.919 billion yuan, a year-on-year increase of 65.22%, and net profit of 135 million yuan, marking a turnaround from losses [8] Group 3: International Expansion - Expanding overseas business is a necessity for Jiadu Technology, with a strategy focused on "deepening domestic and breaking through overseas" [6][7] - The company is preparing for an overseas share issuance (H-shares) and listing on the Hong Kong Stock Exchange to enhance its international competitiveness and brand image [7] - Jiadu Technology plans to establish a research and development center in Hong Kong and a Southeast Asia regional headquarters in Singapore, leveraging technology licensing and capital cooperation [8]

Core Viewpoint - The article discusses the advancements and applications of the "Zhixing Traffic Model" developed by Jiadu Technology, emphasizing its role in transforming traffic management from "intelligent" to "smart" systems through AI and big data integration [3][4][8]. Group 1: AI Strategy and Model Development - Jiadu Technology is focusing on the integration of AI in the transportation sector, with the "Zhixing Traffic Model" being a key innovation that combines over 20 years of industry data and knowledge [4][6]. - The model has evolved from version 1.0 to 2.0, with applications in various scenarios such as customer service, equipment maintenance, video analysis, and passenger flow prediction [6]. - The company aims to create a dual-engine development path with "big model + operating system," planning to release the "Traffic Jiahong" operating system by May 2025 [6][8]. Group 2: Practical Applications and Benefits - The implementation of the Zhixing Traffic Model has led to significant improvements in peak traffic efficiency, such as the use of facial recognition and QR codes for faster subway access [5][6]. - The model facilitates real-time data analysis across different transportation modes, helping to optimize traffic organization and reduce congestion [5][6]. - In Guangzhou, nearly 50 trial intersections have adopted AI-driven traffic signal control, resulting in a 6.5% reduction in congestion during peak hours in key areas [7]. Group 3: Future Goals and Innovations - Jiadu Technology is actively participating in the construction of the Guangzhou Digital Traffic Command System, which integrates 28 types of traffic data and advanced technologies for enhanced traffic management [8]. - The company aims to leverage its AI capabilities to transition traffic systems from reactive to predictive decision-making, enhancing overall traffic governance [8]. - The upcoming 15th National Games will serve as a testing ground for these innovations, showcasing the potential of AI in urban traffic management [8].

Group 1: Company Overview - Keda Technology is focusing on AI as a core strategy, aiming to enhance its competitiveness in the intelligent transportation sector through continuous technological breakthroughs [10][11] - The company has launched the first domestic operating system for the transportation industry, "Traffic Jiahong," to establish a comprehensive and controllable operational foundation [11][12] - Keda Technology's R&D investment for 2024 is projected to be 243 million yuan, with a total of 89 patent applications and 41 granted patents [12] Group 2: International Expansion - Keda Technology views overseas business expansion as a necessity, with plans to establish a new model of "technology output + ecosystem co-construction" starting from Hong Kong [13][14] - The company is preparing for an overseas share issuance (H-shares) and listing on the Hong Kong Stock Exchange to enhance its international competitiveness [14] - Keda Technology aims to leverage its differentiated advantages to penetrate international markets, including establishing a Southeast Asia regional headquarters in Singapore [14] Group 3: Financial Performance - In the first half of the year, Keda Technology achieved a revenue of 4.919 billion yuan, a year-on-year increase of 65.22%, and a net profit of 135 million yuan, marking a turnaround from losses [15] Group 4: Industry Challenges and Strategies - Blue Arrow Electronics emphasizes the importance of continuous innovation to navigate the challenges posed by rising costs and competitive pressures in the semiconductor industry [20][21] - The company is transitioning from consumer electronics to high-value sectors, actively expanding into automotive electronics and industrial applications [21] Group 5: Autonomous Driving Sector - WeRide has been operating safely for over 2,200 days and holds seven national autonomous driving licenses, expanding its operations in 30 cities across 11 countries [22][24] - The company has launched various autonomous driving services, including Robotaxi, and is focusing on international expansion with tailored solutions for different regions [23][24] - WeRide is building an ecosystem of collaboration with major industry players to enhance its competitive edge in the global market [25][26]