Official Amap technology account showcasing all of Amap's technical innovations.
CE-Nav introduces a two‑stage imitation‑then‑reinforcement framework that decouples generic geometric planning from robot‑specific dynamics, enabling low‑cost, high‑performance navigation across quadrupeds, humanoids, and drones while requiring only brief online fine‑tuning.
JanusVLN presents a groundbreaking Vision‑and‑Language Navigation framework that decouples semantic understanding from spatial geometry using dual implicit memory, eliminates explicit memory overhead, achieves state‑of‑the‑art performance with only RGB video input, and dramatically improves efficiency and generalization across VLN benchmarks.
OmniNav introduces a unified framework for embodied navigation that simultaneously handles instruction‑goal, object‑goal, point‑goal, and frontier‑based exploration tasks using a fast visual‑language‑driven policy and a slow memory‑augmented planner, achieving state‑of‑the‑art performance and real‑world 5 Hz deployment.
FantasyHSI introduces a graph‑based multi‑agent framework that combines visual‑language models and video‑generation diffusion to let digital humans perceive, plan, and interact autonomously in any 3D scene, producing physically plausible, long‑duration actions for animation creation and embodied‑AI simulation.
FantasyWorld introduces a geometry‑enhanced framework that augments a frozen video diffusion model with a trainable geometry branch, enabling simultaneous video representation and implicit 3D field generation, achieving spatially consistent, high‑quality virtual worlds and outperforming recent baselines in multi‑view coherence and geometric fidelity.
Gaode Map introduces three AI‑driven routing features—toll‑free exit recommendation, global faster‑route optimization, and high‑frequency rapid‑route detection—that combine massive traffic data, multi‑objective algorithms, and real‑time prediction to help users save both toll costs and travel time during the National Day travel peak.
This article explains how Amap tackles the complex challenge of guiding users safely through diverse intersections by simplifying intersection models, automating topology generation, applying multimodal AI for data updates, and building an AI‑driven navigation agent that delivers accurate, real‑time routing.
FSDrive introduces a spatio‑temporal chain‑of‑thought approach that enables visual language models to generate future driving scenes as images, improving trajectory planning accuracy and safety by eliminating cross‑modal gaps and enforcing physical constraints in autonomous driving.
HS‑STAR introduces a three‑stage hierarchical sampling framework that identifies high‑utility boundary problems, reallocates computation budget to them, and fine‑tunes large language models, achieving significant accuracy gains on math reasoning benchmarks without extra sampling cost.
This article introduces Pos2Distill, a novel knowledge‑distillation framework that transfers capabilities from advantageous to disadvantaged positions in large language models, effectively mitigating positional bias and improving performance on long‑text retrieval and in‑context reasoning tasks.
