0 views collected around this technical thread.
This article explains how Gaode Map leverages lightweight edge TTS models, dual‑autoregressive large‑model data augmentation, and a configurable audio‑processing DAG to enable users to create highly realistic personalized voice packs from just three recorded sentences.
During the 2020 Golden Week travel festival, Gaode Maps deployed a large‑scale serverless architecture using Alibaba Cloud Function Compute to power its core map, route‑planning and navigation services, handling over 100 million users with sub‑60 ms response times, 99.99% success, and dramatically improving development speed, elasticity, cost efficiency and observability.
The 2020 Q2 China Major Cities Traffic Analysis Report, produced by Gaode Maps together with the National Information Center’s Big Data Development Department, Tsinghua University’s Daimler Sustainable Transportation Joint Research Center and other authorities, presents image‑based traffic patterns, mobility trends and data‑driven insights for major Chinese cities in the second quarter of 2020, accessible via the “Read Original” link.
This article details how Gaode Maps leveraged machine‑learning‑based Learning to Rank to rebuild its search‑suggestion ranking pipeline, addressing challenges in sample construction, feature sparsity, and model optimization, and achieving significant improvements in relevance metrics and user experience.
From 2015’s rapid, short‑cycle UI tweaks to a 100‑person front‑end organization, Gaode’s Amap team progressively adopted componentization, templating, a C++‑based dynamic engine and JSX framework, scaling code from 30 k to 600 k lines, and now focuses on fine‑grained, standardized, differentiated, low‑code and platformized development for long‑term stability.
Amap replaced its rule‑based feedback pipeline with a three‑stage, LSTM‑driven system that combines word2vec embeddings and structured fields, achieving over 96% classification accuracy, cutting manual workload by 80%, and slashing per‑task costs while enabling scalable, data‑driven map quality improvements.
Gaode Maps has progressed image segmentation from early heuristic region splitting to modern deep‑learning pipelines—leveraging FCNs, multi‑task networks, Mask R‑CNN, and specialized losses—to achieve centimeter‑level, instance‑aware mapping of roads, signs, and small objects while pursuing lighter, real‑time models.
Gaode Maps applied Learning to Rank to optimize search suggestions, moving from rule-based to gradient boosted rank model, addressing sample construction and feature sparsity via session-based labeling and loss adjustment, achieving a seven‑point MRR gain and higher coverage, and paving the way for personalization and deep learning.