Intelligent Transportation
Search documents
红绿灯背后的AI大模型:从被动响应到全局预判
2 1 Shi Ji Jing Ji Bao Dao· 2025-12-13 10:35
(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 ...
AI如何预判机场大客流、调优红绿灯?这场大会聚焦智慧交通
Nan Fang Du Shi Bao· 2025-12-12 16:51
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' ...
速腾聚创全栈自研芯片通过AEC-Q车规认证,目标识别率提升至99%
Ju Chao Zi Xun· 2025-10-15 03:31
Core Insights - RoboSense has achieved a significant milestone by having its fully self-developed chip pass the AEC-Q automotive certification, making it the only company globally to reach this achievement, which enhances its competitive edge in the lidar technology sector [2][7] Group 1: SPAD-SoC Chip - RoboSense's SPAD-SoC chip is the world's first lidar receiving processing SoC chip and the only one to pass the AEC-Q102 certification, featuring high performance, high integration, and high reliability [3] - The chip improves target recognition rate to 99% and has a sampling capability of 44 billion samples per second, effectively preventing missed detections [3] - It includes a 384-Core signal processing capability for high dynamic response and can detect low-reflectivity objects, while also being resistant to rain, fog, and dust interference [3] Group 2: Proprietary Technologies - The chip incorporates three proprietary technologies that enhance performance and reliability, including Crosstalk Immunity Encoding (ClE), which improves sunlight noise resistance by 95% and cross-talk resistance by 90% [4] - PicoSecond Timing Technology (PTT) enhances distance measurement accuracy by 15 times to 3 cm while reducing power consumption [4] - The Adaptive Fusion Algorithm (AFA) optimizes signal extraction, reducing blind spots by 70 times [4] Group 3: 3D Stacking and Bonding Technology - RoboSense has achieved a global first by integrating SPAD and SoC through 3D stacking, resulting in over 50% reduction in size and high integration [5] - Copper-copper bonding technology provides each SPAD with an independent processing circuit, ensuring low-loss and low-latency signal transmission [5] Group 4: 2D VCSEL Chip - RoboSense's 2D VCSEL chip is the only mass-producible two-dimensional addressable VCSEL chip in the industry, utilizing two-dimensional addressable scanning to suppress high reflectivity expansion [6] - The chip supports flexible matrix scanning, significantly enhancing laser emission efficiency, and features a combined SiP technology that improves accuracy while reducing parasitic parameters [6] Group 5: E Platform - The E platform, equipped with the AEC-Q certified SPAD-SOC and 2D VCSEL chips, is the first mass-producible all-solid-state digital lidar platform in the industry [7] - This platform not only showcases RoboSense's leading position in lidar technology but also provides robust technical support for the development of autonomous driving and intelligent transportation [7]
【前瞻分析】2025年全球智慧交通行业市场规模及投资建设分析
Sou Hu Cai Jing· 2025-09-04 09:10
Group 1 - The global smart transportation market was approximately $96.1 billion in 2022 and is projected to reach $115.8 billion by 2024, with a compound annual growth rate (CAGR) of 8% [1] - The investment in smart city construction globally is expected to approach $80 billion by 2024, driven by urbanization challenges, with Asian countries having significant growth potential [3] - China's fixed asset investment in transportation increased from 2.87 trillion yuan in 2016 to 3.9 trillion yuan in 2023, with a forecast of 3.8 trillion yuan for 2024 [5][6] Group 2 - The market size of smart transportation in China grew from 97.3 billion yuan in 2016 to 236.7 billion yuan in 2023, with expectations to exceed 240 billion yuan in 2024 [7] - The comprehensive three-dimensional transportation network in China has a route mileage exceeding 260,000 kilometers, with a completion rate of about 90% [6]
花旗:千方科技-2025 年第一季度业绩喜忧参半,经营利润未达预期,但投资收益助力盈利超预期
花旗· 2025-05-06 02:28
Investment Rating - The investment rating for China TransInfo Technology is "Sell" with a target price of Rmb5.1, indicating an expected share price return of -39.5% and a total return of -39.2% [3][7]. Core Insights - The company reported a mixed performance in 1Q25, with revenue increasing by 2% year-over-year to Rmb1.6 billion, which was 18% above Bloomberg consensus estimates. However, the operating profit missed expectations, resulting in an operating loss of Rmb77 million [1][2]. - The gross margin expanded by 0.6 percentage points year-over-year to 29.7%, but this was still 4.6 percentage points below consensus estimates. Operating expenses were flat year-over-year at Rmb555 million, which was 34% higher than expected [1][2]. - Net profit for 1Q25 was Rmb221 million, significantly exceeding consensus estimates by 101%, primarily due to a gain from a fair value change in investments amounting to Rmb302 million [1][2]. Summary by Sections Financial Performance - 1Q25 revenue was Rmb1.609 billion, 17.9% higher than consensus estimates and 1.8% higher year-over-year. Gross profit was Rmb478 million, reflecting a 4% increase year-over-year [2][5]. - The operating loss of Rmb77 million resulted in an operating margin of -4.8%, which was 8.7 percentage points below consensus expectations [2][5]. - The net income of Rmb221 million marked a significant recovery from a net loss of Rmb58 million in the same quarter last year [2][5]. Valuation - The target price of Rmb5.1 is based on a price-to-earnings ratio of 19.0x for the second half of 2024 to the first half of 2025, aligning with the five-year sector average [7].
Actelis Networks Receives Follow-On Order from City of Eugene, Oregon for Traffic Management System Modernization
Newsfilter· 2025-04-14 12:30
Core Viewpoint - Actelis Networks, Inc. has received a follow-on order from the City of Eugene, Oregon to expand its traffic management system, indicating the growing adoption of its hybrid fiber-copper networking solutions for smart city applications [1][2][5] Company Overview - Actelis Networks is a market leader in cyber-hardened, rapid deployment networking solutions, specializing in hybrid fiber-copper technologies for IoT and broadband applications [1][5] - The company offers fiber-grade performance with the flexibility and cost-efficiency of hybrid fiber-copper networks, enhancing connectivity for various sectors including government and transportation [5] Recent Developments - The new order builds on a previous deployment announced in February 2023, as part of Eugene's Transportation Management Plan aimed at tripling the current rate of transit by foot, bicycles, and public transport by 2035 [2] - Actelis' technology integrates with traffic controllers and IoT devices, providing live data to the city's operations center, thereby modernizing traffic management capabilities [3][4] Partnerships - The equipment is being delivered through Western Systems, a long-standing partner of Actelis, which collaborates with various agencies to implement advanced traffic solutions across the western United States [4]