The China Graduate Electronic Design Contest, initiated in 1996 by the China Electronics Society and Tsinghua University, has been successfully held for twenty-eight years. The competition has always adhered to the purpose of "inspiring innovation, encouraging entrepreneurship, improving quality, and strengthening practice." After more than 20 years of development, the competition has become an effective platform for graduate students in electronic-related disciplines to cultivate innovation and entrepreneurship awareness, enhance team spirit, improve practical abilities, and promote the comprehensive growth of young innovative and entrepreneurial talents.

Topic 1: Development of an Edge Image Processing System Based on TPU Processor

Background of the Topic

The advancement of Internet of Things (IoT) technology has promoted the development of edge computing, a computing paradigm that moves data processing from centralized cloud to the network edge closer to the data source. This reduces latency, enhances data privacy, and reduces reliance on cloud computing. Image processing is a key task in edge computing; devices such as smart cameras and drones need to process images in real-time to perform functions like object detection and face recognition. However, the limited computational resources of devices often lead to slow processing speeds and high power consumption. To solve these problems, specialized processors such as Tensor Processing Units (TPUs), which are computationally efficient and consume less energy when performing intelligent tasks, have been designed. Utilizing these processors, edge devices can achieve fast and energy-efficient image processing, improving performance and efficiency. Therefore, the development of edge image processing systems based on TPU processors has become a hot research topic, capable of meeting the demands of real-time image processing and providing strong support for applications in smart security, transportation, and retail.

Task of the Topic

This topic aims to use the Shaolin development board to create an intelligent system, covering software and hardware design, algorithm innovation, and new application development, mainly for consumer electronics and robotics. This system will enhance the interactive experience of smart devices and is suitable for environmental perception and interaction in smart homes, robots, etc.

1. Integrated software and hardware design: Design an efficient edge image processing system based on the TPU processor to improve data processing speed and energy efficiency, ensuring the collaborative work of software and hardware.

2. Algorithm innovation: Develop efficient image and voice recognition algorithms, as well as other interactive technologies, to enhance the interactive capabilities of consumer electronics and robots.

3. Multi-scenario application development: Develop environmental perception and interaction applications for consumer electronics or robots, achieving multimodal interaction.

4. System testing and validation: Test newly developed applications in real environments to ensure their stability and reliability, meeting practical needs.

5. Technology promotion and commercialization: Transform technological achievements into commercial products, promote together with partners, and facilitate the popularization of intelligent applications.

Through the combination of software and hardware, algorithm optimization, and actual application development, we aim to create an innovative intelligent system that drives the innovation of consumer electronics and robotics technology.

Topic 2: Application System Development Based on Airbox for Large Language Models

Background of the Topic

In the field of deep learning, large language models have become a major force driving technological progress. These models have shown remarkable capabilities in text understanding, generation, translation, summary, question-answering, and other aspects. Airbox edge computing box provides an efficient, low-latency platform for these models, allowing large language models to operate closer to the data source, offering faster responses and higher privacy protection. This topic aims to explore the application potential of large language models in different industries and scenarios.

Task of the Topic

Participants need to choose a specific application scenario and design and develop an application system based on the Airbox edge computing box. The system should integrate large language models to solve practical problems or needs in that scenario. Through Airbox, we can not only realize the optimization and application of large language models in traditional cloud computing environments but also apply them to embedded scenarios such as Network Attached Storage (NAS), home gateways, or other AIoT devices. The system must meet the following requirements:

1. Performance optimization: The system should leverage Airbox's low latency and local computing capabilities to optimize the performance of the language model, ensuring fast and accurate responses.

2. Privacy and security: The system should ensure privacy and security in processing and storing user data, complying with relevant regulations.

3. Scalability and maintainability: The system design should be scalable and maintainable to facilitate future updates based on changing requirements.

Topic 3: Application System Development for Empowering Traditional Industries with Intelligent Technology Based on CV1800B or SG200x

Background of the Topic

Against the backdrop of intelligent technology continuously driving the transformation of traditional industries, the emergence of TPU and RISC-V architecture dual-core processors CV1800B and SG200x has provided strong hardware support for industry intelligence. Development boards based on these processors, with their high-performance computing and efficient energy ratio, bring new opportunities for intelligent integration in manufacturing, retail, medical, and other industries.

This topic focuses on using the technical characteristics of CV1800B and SG200x processors to develop innovative application systems to meet the needs of specific industry scenarios. Contest entries need to demonstrate the system's efficient performance, stability, and compliance in terms of security and scalability. The competition aims to reward solutions that can achieve breakthroughs in practical applications and promote the deep integration of intelligent technology with traditional industries.

Task of the Topic

Participants need to use the CV1800B/SG200x series development board to integrate algorithms with traditional microcontroller-based industry applications to develop innovative intelligent application systems. The task direction includes but is not limited to areas such as agricultural intelligence, robotics, etc. These systems should demonstrate the following capabilities for the selected traditional industry scenarios:

1. Performance optimization: Fully utilize the high-performance computing power of the processor to ensure rapid response and efficient data processing of the application system.

2. Scenario adaptability: Customize development functions according to the needs of specific industries to enhance the level of intelligence of the scenario.

3. System stability: Ensure the system operates stably in real industrial environments, including fault recovery and long-term operational reliability.

Task Objectives:

1. Design and implement one or more solutions that effectively address the specific challenges of the selected industry.

2. Provide comprehensive system design documentation, covering architecture, functionality, and performance indicators.

3. Demonstrate the actual operation of the system and verify its efficiency and stability through test cases.

4. Analyze the business potential of the solution, its feasibility of implementation, and its long-term positive impact on the industry.

The 19th China Graduate Electronic Design Contest is divided into preliminary and final rounds, with the preliminary round being held across eight national regions: Northeast, North China, Northwest, Central China, East China, Shanghai, Southwest, and South China. The regional contests are hosted by Northeastern University, Shandong University, Xi'an University of Technology, Hubei University, Zhejiang University of Technology, Shanghai Society of Electronics, University of Shanghai for Science and Technology, Chongqing Jiaotong University, and Guilin University of Electronic Technology, respectively.

• Start of the competition: March 2024
• Registration deadline: June 20, 2024
• Preliminary and final rounds: The preliminary round is expected to be held in early July, and the national finals are scheduled for mid-August.

China Graduate Electronic Design Contest· SOPHGO CUP Official QQ Discussion Group: 160584984

1. Graduate students currently enrolled in mainland China, Hong Kong, Macau, and Taiwan (including recent graduates), senior undergraduates who have been granted graduate admission (with proof of school recommendation or admission), and international university graduate students are eligible to participate in the contest.
2. Teams, rather than individuals, must register to compete.
3. The competition is divided into two main parts: a technical contest and a business plan competition, each conducted independently with separate team formation, registration, evaluation, and awards, without affecting each other.
4. The technical contest consists of enterprise-set problems and open problems. Teams participating in the enterprise-set track will be judged by the proposing enterprises and will receive awards for both the preliminary and final rounds of the technical contest, as well as special enterprise awards. These awards are independent and complementary, with the higher prize money being awarded.
5. Each team consists of a mentor, a team leader, and members. Within the same competition section (technical contest or business plan competition), there should be one or two mentors, one team leader, and one to two members. Each mentor can guide up to two teams, and each team member can only join one team.
6. Participants must have official graduate student status or confirmed admission qualifications. In special circumstances, a maximum of one undergraduate student may be allowed in a team, but this must be applied for in advance to the competition organizing committee and approved. Details for undergraduate applications can be found in the official website's annex.
7. The registration fee for both the technical contest and the business plan competition is collected by the organizer and the secretariat, the China Electronics Society, which will issue an electronic invoice. The registration fee is 500 RMB per team in principle, but special circumstances may allow for adjustments by the regional competitions, which will be officially notified.
8. During technical contest registration, teams are required to fill in a project summary that describes the innovative aspects of their work, including three key innovation keywords and an explanation of member roles.
9. Before the registration deadline, team member information, member order, and project information filled out on the competition website can be changed. After the registration deadline, team members and the project cannot be changed. After entering the finals, teams may improve and perfect their projects, but cannot change the content of the project or team members. If changes are necessary due to special circumstances, teams must submit a change request to the organizing committee, which can only be made with approval. Unauthorized changes to competition information may result in disqualification by the organizing committee.