Jinli Technology

Committed to becoming a core switch service provider

Artificial intelligence and high-performance computing

Edge computing and the Internet of Things

Intelligent Network

Data storage and management

Partner Joint Program

Autonomous driving solutions

An autonomous driving solution jointly developed with ecological partners. It provides the core driving force for autonomous driving model training through leading computing infrastructure, data management, data annotation, and model training. It efficiently builds a full-stack cluster solution from "data acquisition, storage, and processing" to "model training" and "simulation verification", solving the problems of insufficient computing power, insufficient storage resources, and difficult-to-use software for autonomous driving customers.

Scenario Analysis

The development of autonomous driving models requires a large amount of data collection and processing. Processed data is then used for model development and training. This entire workflow requires substantial computing resources and a rich basic software environment.

Solution

A joint solution developed with ecological partners. It leverages leading computing infrastructure, data management, data annotation, and model training to provide the core driving force for autonomous driving model training. It efficiently builds a full-stack cluster solution from "data acquisition, storage, and processing" to "model training" and "simulation verification," addressing the issues of insufficient computing power, insufficient storage resources, and difficult-to-use software for autonomous driving customers.

The solution mainly includes AI servers and CPU servers, using an AI Pod networking method, along with modules such as the AIStation artificial intelligence development platform:

1. AI servers utilize a leading technology architecture to provide strong computing power for the algorithm development of autonomous driving single-task models and fusion models;

2. CPU servers provide stable and reliable computing support for data management, data annotation, and big data management;

3. The AIStation artificial intelligence development platform unifies the management of AI applications and provides users with algorithm platform and application optimization services;

4. AI Pod networking is specifically designed for fusion models, providing effective network communication guarantees for strong AI computing power.

Application Scenarios

The solution provides basic computing power and software platform support for autonomous driving model development, offering full-stack computing power and software platform support from data acquisition, data annotation, data management, model training, and simulation verification.

Solution Value


Platform optimization for fusion models	A complete Yuannao ecosystem	A powerful resource management platform	AI Pod networking	Globally leading heterogeneous computing architecture
Based on the autonomous driving solution as a basic platform, a spatiotemporal fusion model architecture based on multiple cameras has been developed, significantly optimizing the speed of target object monitoring and displacement direction prediction. It topped the leaderboard for the nuScenes pure vision 3D object detection task in the autonomous driving dataset, raising the key indicator nuScenes Detection Score (NDS) to 62.4%.	The Yuannao ecosystem, in collaboration with left-hand partners possessing algorithm development capabilities and right-hand partners with extensive industry implementation experience, jointly provides data management and data processing platform support for autonomous driving users.	Provides reliable, easy-to-use, and flexible service deployment and computing resource management platforms for autonomous driving algorithm R&D users, helping users quickly launch their businesses and improve the utilization efficiency of computing resources.	A leading AI Pod networking method, specifically designed for massive models, effectively assists autonomous driving customers in iteratively upgrading from single-task models to fusion models.	The AI server combines the graphics computing, AI computing, high-speed storage, and low-latency network of high-performance GPUs, achieving a processing speed of 13,000 images per second, providing strong computing power for autonomous driving model development.