How Didi Maps Engineered Scalable AR Navigation for Airports and Malls

Background

Large indoor venues such as airports, train stations, and shopping malls suffer from weak GPS signals, extensive area, and complex routes, which increase driver‑passenger meeting cost and reduce efficiency.

Technical Objectives

The system addresses three fundamental questions—where am I, where do I want to go, and how do I get there—by focusing on mapping, localization, and navigation planning.

Mapping (3D Reconstruction)

Continuous‑frame optimization and data‑association strategies were introduced to overcome the size limitation of single‑camera visual modeling. This enabled the construction of indoor maps covering more than 60,000 m² with high precision.

Localization (Monocular Visual Positioning)

A monocular visual positioning pipeline based on image matching and visual‑odometry was developed. Innovations in data retrieval reduced the average positioning error to less than 0.5 m, roughly 1/30 of typical GPS error, outperforming Wi‑Fi, Bluetooth, and UWB solutions.

Navigation Planning and Interaction

Inertial sensor data are fused to render AR arrows and virtual road signs that guide users to recommended pick‑up points. The Android implementation leverages Google ARCore for real‑time visual rendering and interaction.

Deployment and Device Support

The AR navigation service is live in several Chinese airports, train stations, and large malls. Initial device support includes iPhone 7 and newer iOS devices, as well as Huawei P20 and P20 Pro, with plans to expand to additional models.

Impact

Since rollout, the solution has reduced driver‑passenger meeting time and improved navigation efficiency in complex indoor environments.

Code example

推荐阅读
▬
更多推荐
▬
滴滴开源
/ Open Source
Levin
|
AoE
|
Delta
|
Mpx
|
Booster
|
Chameleon
|
DDMQ
|
DroidAssist
|
Rdebug
|
Doraemonkit
|
Kemon
|
Mand Moblie
|
virtualApk
|
获取更多项目
技术干货
/ Recommended article
重磅！滴滴跨端框架Chameleon 1.0正式发布
|
Android 性能优化之 Activity 启动耗时分析
|
HDFS 源码解读：HadoopRPC 实现细节的探究
|
阅读更多内容