类脑智能应用技术研究中心致力于探索类脑智能前沿技术,目标是在机器上实现类人水平的感知、推理和决策能力。为实现该目标,实验室在相关的基础理论、智能算法和智能硬件系统方面开展研究,主要研究方向包括:
(1)类脑算法:研究类脑智能的计算模型与学习机制,重点解决传统算法在能效、自适应性和泛化能力上的不足。核心包括脉冲神经网络(SNN)、类脑强化学习、神经形态计算等,模拟生物神经系统的动态特性与可塑性,提升复杂场景下的实时学习与决策效率。
(2)类脑芯片:设计支持类脑算法的专用硬件,突破传统架构在并行计算与能效上的瓶颈。研究神经形态芯片,通过存算一体、事件驱动等技术实现低功耗、高并发的脉冲信号处理,为边缘计算与实时应用提供硬件基础。
(3)类脑系统:构建软硬协同的类脑智能系统,解决类脑算法与芯片系统集成与规模化部署问题。研究内容包括类脑计算框架、类脑机器人系统等,优化算法-芯片-传感器的协同,实现感知-决策-控制闭环,推动自动驾驶、智能制造与智能医疗等应用落地。
类脑智能应用技术研究中心,隶属于上海交通大学集成电路学院,坐落在上海交通大学闵行校区微电子楼,拥有价值2000多万元的实验设备,实验室配备了丰富的计算资源,包括数台高性能多卡GPU服务器,可满足AI研究高计算需求,具有良好的研究环境和学术氛围。中心现有教授、副教授、研究员、副研究员、高级工程师、博士后、博士生、硕士生和工程师50多人。中心在类脑计算、类脑芯片架构、3D视觉智能处理、融合高精度导航、基于语义拓扑网的类脑多目标感知与导航等方面取得了丰硕成果,获得国家、地方政府与企业的大力支持,拥有专利百余项,发表论文百余篇,获国家科技进步二等奖1项,上海市科技进步特等奖1项。
The brain-inspired application technology center (BATC) is committed to exploring the frontier technology of brain-like intelligence, with the goal of realizing human-like perception, reasoning and decision-making ability on machine. For this end, BATC has conducted many researches on the relevant basic theories, intelligent algorithms and hardware systems. The main contents include:
(1) Brain-inspired Algorithms: Research on computational models and learning mechanisms of brain-inspired intelligence, focusing on addressing the shortcomings of traditional algorithms in energy efficiency, adaptability, and generalization capabilities. The core areas include Spiking Neural Networks (SNN), brain-inspired reinforcement learning, neuromorphic computing, etc., which simulate the dynamic characteristics and plasticity of biological nervous systems to enhance real-time learning and decision-making efficiency in complex scenarios.
(2) Neuromorphic Chip: Design dedicated hardware supporting brain-inspired algorithms to break through the bottlenecks of traditional architectures in parallel computing and energy efficiency. Research neuromorphic chips, which achieve low-power and high-concurrency processing of spike signals through technologies such as in-memory computing and event-driven mechanisms, providing a hardware foundation for edge computing and real-time applications.
(3) Brain-inspired System: Build a software-hardware collaborative brain-inspired intelligence system to address the integration and large-scale deployment of brain-inspired algorithms and chip systems. Research areas include brain-inspired computing frameworks, brain-inspired robotics systems, etc., optimizing the coordination of algorithms, chips, and sensors to achieve perception-decision-control closed loops, and promoting the implementation of applications such as autonomous driving, intelligent manufacturing, and smart healthcare.
BATC is subordinate to the School of Electronic Information and Electrical Engineering in Shanghai Jiao Tong University, located in the Microelectronics Building of Minhang District. It has experimental equipment worth more than 20 million CNY, our laboratory is equipped with abundant computing resources, including several high-performance multi-GPU servers, which can meet the high computational demands of AI research. There is a good research environment and academic atmosphere here. At present, there are more than 50 people in BATC, including professor, associate professors, researchers, senior engineer, post doctor, postgraduate students and engineers. BATC has made fruitful achievements in brain-like computing and neuromorphic architecture, 3D visual intelligent processing, integrated high-precision navigation, brain-like multi-target perception and navigation based on semantic topology network, etc. It has obtained the strong support of national and local governments and enterprises, applied for more than 100 patents and published more than 100 papers. More importantly, BATC won a second prize of National Science and Technology Progress, and a special prize of Shanghai Science and Technology Progress.