Gaode Traffic Light Countdown Goes Global
Quick Look
- Gaode's AI-powered traffic light countdown feature, launched in 2022, is expanding globally.
- The technology, which simulates and calculates traffic signal changes without direct hardware access, has covered nearly 500,000 intersections in China and is now set for international rollout.
AI-generated summary
Why It Matters
Gaode's traffic light countdown feature, launched in May 2022, utilizes AI to simulate and calculate traffic signal changes, providing real-time countdowns and waiting turn information to users. It has been implemented across nearly 500,000 intersections in mainland China, Hong Kong, Macau, and Taiwan, enhancing traffic efficiency and safety.
5月20日,高德红绿灯倒计时宣布出海,将在全球更多国家和地区逐步落地。
红绿灯倒计时是阿里巴巴旗下高德于2022年5月推出的产品功能,基于AI技术,能够对路口信号灯变化进行实时模拟与计算。
在实际使用中,用户无论是在驾车、骑行、步行等导航状态下,还是巡航、驾驶模式中,经过支持该功能的路口时,均可在高德App看到红灯倒计时读秒;当路口车流较大,高德App还会提示需要等待的红灯轮次。
截至目前,高德红绿灯倒计时已在中国大陆及港澳台地区全面落地,覆盖红绿灯路口近50万个,有效提升了交通路口的通行效率与安全性。
在全球导航市场,信号灯倒计时是一项公认的技术难题。由于各国交通信号系统标准各异、设备数据开放程度不同,海外主流导航产品均未提供大规模的信号灯精确倒计时服务。
而高德首创的红绿灯预测技术是通过AI大模型对历史交通数据、实时车流信息与路口信号规律进行综合计算,从而实现不依赖信号灯硬件接入的倒计时模拟。
目前,高德红绿灯预测技术已完成多次技术迭代。基于空间智能推出的视觉时空序列模型(VSTM),在时空计算中引入视觉时序感知,使系统能够直接洞察路口实时动态,不再依赖传统时空序列特征,从而更准确地理解和预测复杂路口的交通变化。
技术储备上,高德早在2019年4月就开始申请红绿灯倒计时相关专利,是行业最早进行相关技术布局的企业之一。目前,高德已成功申请多项技术专利,覆盖交通数据的周期挖掘、相位计算及前端导航体验优化等核心环节,形成了从后端算法到前端体验的全栈技术布局。
在完成中国全域覆盖后,高德计划将这套经过充分验证的AI技术方案推向全球更多国家和地区。未来,高德用户在海外也将能够体验到红绿灯倒计时带来的出行便利。
What to Watch
AI outlook — possibilities, not facts
Gaode's traffic light countdown feature will be implemented in multiple countries outside of China within the next year.
Likely · Within months
Open Questions
- Which specific countries or regions will be the first to implement the feature globally?
- What are the specific challenges Gaode anticipates in adapting its technology to different international traffic systems?
- Will there be any partnerships with local navigation or automotive companies in the global rollout?
- What is the expected timeline for the global rollout beyond the initial phase?





