AIRS needed scalable compute, storage and scheduling infrastructure to tackle the algorithm, data and compute bottlenecks that were limiting embodied-intelligence research and deployment.
Reported impact
AIRS gained a cloud-scale embodied-AI research base that improved compute utilization, supported end-to-end experimentation and accelerated the path from frontier research to industry deployment.
AIRS built its embodied-intelligence research environment on Baidu AI Cloud's infrastructure layer so it could combine model, data, compute and platform services in one reproducible stack. The implementation integrates GPU cloud servers, the BaiGe AI heterogeneous-compute platform, high-performance storage and Qianfan-related model services into a mixed resource pool that can support training, simulation and real-time inference workloads at the same time. That matters because AIRS is advancing several open systems: AIRSHIP for embodied systems, AIRSPEED for multimodal data generation and labeling, and AIRSTONE for low-power embodied compute. Baidu's contribution is not just raw hardware capacity; it also includes heterogenous resource scheduling, model compression support, edge-network optimization for end-cloud coordination and a parallel file system capable of storing the large sensor and training datasets generated by robotics research. The resulting environment gives AIRS a shared cloud base for embodied-model experimentation, data synthesis and deployment-oriented algorithm work.
Training data, simulation outputs and real-time inference workloads are scheduled through BaiGe across CPU and GPU clusters; the resulting datasets are stored in parallel file storage and fed back into model training, evaluation and embodied-control research on the HPC platform.
AIRS faced the three core embodied-intelligence challenges of algorithms, compute and data (original: 在中游技术方面具身智能面临算法、算力和数据三重挑战).
Baidu AI Cloud built a mixed resource pool with multiple GPU specifications, CPU clusters and high-performance storage to support training, data synthesis and real-time inference (original: 整合多种性能规格GPU……CPU 集群及高性能存储，构建 “通用算力 + 专用加速” 混合资源池).
Parallel high-performance storage ensures that massive embodied-intelligence datasets can be stored and read efficiently (original: 高性能的并行文件存储，则确保具身智能研发中产生的海量数据得以安全、高效存储和读取).
The optimization plan reduces end-cloud coordination delay to within 50 milliseconds to meet real-time robot-control requirements (original: 同时通过优化边缘节点网络，减少端云协同延时（50ms以内），满足具身机器人实时控制要求).
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Baidu AI Cloud Sample Oct 12, 2025 Guiyang Public Security Traffic Management Bureau (贵阳市公安交通管理局) Guiyang needed a more systematic way to reduce congestion in a mountainous city where road geometry, rapid growth, and fragmented traffic-management methods made it hard to optimize signal timing and respond to changing… Average vehicle delay at target intersections fell 15%. AI Traffic Signal Control Large Model · Qianfan Application Development Platform · Qianfan Model Development Platform Baidu AI Cloud Sample Sep 24, 2025 All-China Federation of Trade Unions The All-China Federation of Trade Unions needed a unified AI foundation that could support legal aid, rights protection, and other worker services at national scale instead of leaving union service digitization… Delivery cycle fell from 3-6 months to under 1 month. Qianfan AppBuilder · Qianfan ModelBuilder · Union Industry Model Baidu AI Cloud Sample Jun 19, 2025 Haikou Traffic Police Detachment After Typhoon Yagi damaged most field sensing equipment, Haikou traffic police urgently needed to restore signal-control effectiveness without waiting for every radar, camera, and detector to be rebuilt, or the city… 567 intersections were affected by the typhoon. Edge Computing Node · Qianfan AppBuilder · Qianfan ModelBuilder 