Projects科研项目
Applied platforms built around Ubiquitous Psychological Computing and Embodied Emotional Intelligence.围绕普适心理计算与具身情感智能两大框架，面向产业、公共服务与重大场景打造可落地的情感智能和人因智能系统方案。
Selected National and Major Projects代表性国家级与重大项目
The lab's research is supported by a focused portfolio of national, provincial, and mission-oriented projects. The list below is intentionally selective: it highlights project level, technical direction, and platform value rather than every administrative detail.团队长期承担国家级、省部级和任务牵引型科研项目。这里选择性展示项目层级、技术方向和平台价值，不把网页做成完整项目账本。
Affective Perception and Cognitive Regulation for Large Models大模型情感感知与认知调控
A National Natural Science Foundation of China project exploring how large models can perceive emotion, represent mental state, and support cognitive regulation in emotionally complex human-AI interaction.国家自然科学基金项目，面向大模型情感感知、心理状态表征与认知调控，支撑具身情感智能和情感支持交互研究。
AI for Healthy Living Environment Measurement, Evaluation, and Control人工智能健康居住环境测-评-控
A National Key R&D Program task connecting affective computing, environmental sensing, human factors, and active health into intelligent products for healthy living environments.国家重点研发计划任务，将情感计算、环境感知、人因工程和主动健康连接起来，服务健康人居环境中的智能测评与控制产品。
Multi-Role Multimodal Interaction for Affective Guidance面向情感引导的多角色多模态交互
An NSFC General Program project on multimodal perception, affective understanding, and interaction modeling for multi-role emotional guidance scenarios.国家自然科学基金面上项目，研究多角色场景下多模态情感感知、情感理解与情感引导交互建模。
Non-Contact Physiological-Psychological Correlation Modeling非接触式生理心理关联模型
A national task developing non-contact sensing and physiological-psychological correlation models, forming an important technical foundation for ubiquitous psychological computing.国家重点研发计划相关任务，围绕非接触感知和生理-心理关联建模展开，是普适心理计算框架的重要技术基础。
AI-Assisted Autism Assessment and Intervention儿童孤独症辅助诊断与干预
A key R&D project integrating human-computer interaction, behavioral sensing, and intelligent intervention to support child development and autism-related assessment scenarios.安徽省重点研发项目，融合人机交互、行为感知与智能干预，服务儿童孤独症辅助诊断、评估和干预场景。
Psychological Quantification and Decision Support in Special Scenarios特殊场景心理量化与决策支持
Mission-oriented and public-safety-related projects extend the lab's methods to high-stakes scenarios where emotion, stress, cognitive load, and decision behavior must be modeled responsibly.面向公共安全、国防和特殊任务场景的心理量化、动态情感认知与决策支持研究，强调高可靠场景中的负责任建模与工程落地。
From Lab Prototype to Industrial Deployment从实验室样机到产业化部署
MAC-Lab's distinctive value is not only that it can publish or build a platform. The lab has worked through the difficult middle where a research result must face data uncertainty, hardware constraints, user behavior, privacy boundaries, product definition, field validation, and partner coordination.MAC-Lab 的特点不只是能发论文，也不只是能做平台。团队真正反复经历过的是科研成果走向真实应用时最难的中间地带：数据不确定性、硬件约束、用户行为、隐私边界、产品定义、场景验证和合作方协同。
Start from questions that can support high-level research从能支撑高水平研究的问题出发
Projects are connected to long-term questions in emotion, cognition, psychological state, multimodal sensing, and embodied interaction.项目与情绪、认知、心理状态、多模态感知和具身交互中的长期科学问题相连接，避免只做一次性工程拼装。
Build the chain from model to device to platform打通模型、设备与平台链条
The lab works across sensing modules, affective models, interaction design, edge deployment, user interfaces, and platform-level management.团队跨越感知模块、情感模型、交互设计、边缘部署、用户界面和平台化管理，不把算法孤立在论文或演示里。
Test with real people, settings, and constraints在真实人群、真实环境和真实约束中验证
Psychological computing must be evaluated under realistic interaction, sensor noise, compliance limits, and user trust conditions.心理计算必须在真实交互、传感噪声、合规限制和用户信任条件下验证，才能从“技术可行”走向“场景可用”。
Leave partners with reusable capability, not a one-off demo给合作方留下可复用能力，而不是一次性演示
Good projects should produce reusable assets: methods, patents, standards awareness, datasets, evaluation protocols, student talent, and deployable modules.好的项目应沉淀方法、专利、标准意识、数据、评测流程、学生人才和可部署模块，而不是停留在一次性展示。
Industry Solution Portfolio面向行业的解决方案体系
Affective and human-factors problems usually sit inside messy real scenarios. A partner may begin by asking for emotion recognition, but the real need often involves state sensing, human-factor interpretation, mind-body capability assessment, cognitive-load modeling, intervention design, device integration, management workflow, and deployment evaluation. MAC-Lab can join at any layer, but its strongest value appears when these layers are designed together.行业中的情感与人因问题往往嵌在复杂场景里。合作方一开始可能提出“情绪识别”，但真实需求常常同时包含状态感知、人因解释、身心能力评价、认知负荷建模、干预设计、设备集成、管理流程和部署评估。MAC-Lab 可以从任一层切入，但更大的价值在于把这些层一起设计成完整方案。
Multimodal affective perception that fits the scene贴近场景的多模态情感感知
Emotion recognition is treated as a technical component, not the whole answer. Depending on the scene, the route may combine affective NLP, speech emotion, facial expression, posture, gaze, physiological signals, environmental variables, and interaction behavior.情绪识别只是技术组件，不是完整答案。根据场景不同，方案可融合情感 NLP、语音情绪、面部表情、姿态、眼动、生理信号、环境变量和交互行为。
Models for stress, fatigue, attention, and cognitive load压力、疲劳、注意与认知负荷建模
Industry partners need to know not only what a person feels, but what that state means for safety, decision making, learning, service quality, health risk, or job performance.产业伙伴需要知道的不只是“人是什么情绪”，还要知道这种状态对安全、决策、学习、服务质量、健康风险或岗位绩效意味着什么。
Mind-body capability and scenario-specific evaluation身心能力与场景化评价
The lab can help design assessment indicators, baselines, protocols, dashboards, and validation methods for role fitness, training effect, psychological risk, resilience, and human-performance optimization.可围绕岗位适配、训练效果、心理风险、韧性和身心效能优化，设计评价指标、基线、流程、看板和验证方法。
Software-hardware platforms that can be tested in the field可现场验证的软硬件平台
Deliverables can include algorithm services, assessment tools, sensing terminals, intervention cabins, smart cockpit modules, digital-human interfaces, management platforms, and pilot validation reports.交付形态可包括算法服务、评测工具、感知终端、干预座舱、智能座舱模块、数字人界面、管理平台和试点验证报告。
The hard part is not one model. It is the whole route.难点不是某一个模型，而是整条路线。
In real projects, affective computing must answer several questions at once: what signal is reliable, what state should be modeled, what decision is allowed, what feedback is safe, what hardware is practical, what data can be collected, and how the system will be maintained. MAC-Lab has accumulated experience across this full route, from papers and patents to standards, devices, platforms, and field scenarios.真实项目中的情感计算，必须同时回答一组问题：什么信号可靠，应该建模什么状态，哪些决策可以支持，什么反馈是安全的，硬件是否现实，数据能否持续采集，系统如何维护。MAC-Lab 的积累覆盖了这条完整路线，从论文、专利到标准、设备、平台和现场场景。
Start from a scenario, not from a buzzword.从场景开始，而不是从概念开始。
Typical collaboration can begin with a field diagnosis, technical route design, pilot dataset, algorithm module, evaluation protocol, prototype platform, or joint research project. The goal is to leave partners with a clearer product direction and reusable capability, not only a short-term demonstration.合作可以从场景诊断、技术路线设计、试点数据集、算法模块、评测方案、原型平台或联合课题开始。目标是帮助合作方形成更清晰的产品方向和可复用能力，而不是只完成一次短期演示。
Frameworks and Applied Platforms框架与应用平台
Ubiquitous Psychological Computing Framework普适心理计算框架
- Focus研究焦点
- Continuous psychological modeling, non-contact sensing, dynamic profiling, risk warning, individualized intervention连续心理建模、非接触感知、动态画像、风险预警与个体化干预
- Application应用场景
- Smart cockpit, education, healthcare, healthy living, public safety, human-performance support智能座舱、教育、医疗健康、健康人居、公共安全与人因效能支持
This is MAC-Lab's core framework for moving affective computing into everyday environments. It models emotion and psychological state continuously across time, context, individual difference, physiological-behavioral signals, and group-level indicators, then connects sensing, assessment, warning, and intervention into one technical route.这是 MAC-Lab 推动情感计算进入真实生活环境的核心框架。它从时间、环境、个体差异、生理行为信号与群体指标等维度对情绪与心理状态进行连续建模，并把感知、评估、预警与干预连接成一条完整技术路线。
The framework is designed for scenes where psychological state cannot be understood from a single utterance or a single image. It treats emotion as a temporal and situated process, using multimodal sensing, behavioral context, and human feedback to support more responsible assessment and intervention.该框架面向那些无法依靠一句话或一张图理解心理状态的场景。它把情绪看作具有时间性和情境性的过程，利用多模态感知、行为情境和人类反馈，支持更负责任的评估与干预。
emotion, fatigue, stress, and cognitive-load monitoring for safer mobility.面向出行安全的情绪、疲劳、压力和认知负荷监测。
environment-human state coupling for active health and living-space intelligence.连接环境状态与人的身心状态，服务主动健康与人居智能。
screening, early warning, and supportive feedback for students and special groups.为学生和特殊群体提供筛查、预警与支持性反馈。
Embodied Emotional Intelligence Framework具身情感智能框架
- Focus研究焦点
- Embodied emotional interaction models, multimodal affective perception, expressive generation, situated interaction具身情感交互模型、多模态情感感知、情感表达生成与场景化交互
- Application应用场景
- Companion robots, digital life partners, eldercare, medical rehabilitation, education, intelligent vehicles陪伴机器人、数字生活伙伴、养老照护、医疗康复、教育与智能车辆
This framework gives embodied AI an emotional interaction layer. Built around embodied emotional interaction models and modular emotional capabilities, it supports the full chain of emotional perception, emotional understanding, emotional expression, and emotional interaction for robots, digital humans, smart devices, and active-health systems.该框架为具身 AI 增加情感交互层，以具身情感交互大模型和模块化情感能力为核心，覆盖情感感知、情感理解、情感表达与情感交互全链路，面向机器人、数字人、智能终端和主动健康系统提供“情感大脑”。
The key is not to make a device appear emotional, but to make its response appropriate to the situation. That requires multimodal perception, social timing, expressive control, memory, safety boundaries, and the ability to act with restraint in sensitive contexts.关键不是让设备“看起来有情绪”，而是让它在具体情境中的回应更合适。这需要多模态感知、社会时机理解、表达控制、长期记忆、安全边界，以及在敏感场景中克制行动的能力。
Brain情感
大脑 Perceive感知 Understand理解 Express表达 Interact交互 Remember记忆 Stay Safe安全
emotional perception and situated support for companionship, care, and service.面向陪伴、照护和服务的情感感知与场景化支持。
voice, expression, personality, and long-term interaction memory.融合语音、表情、人格和长期交互记忆。
emotion-aware interaction for rehabilitation, eldercare, and mental support.服务康复、养老和心理支持的情感化交互。
MindOS
MindOS is our affective mind-computing foundation. It is positioned as a research route for psychological digital twins, long-term embodied agents, and scenario reasoning from individuals to groups.MindOS 是我们的情感心智计算底座，定位为支撑心理数字孪生、长期具身智能体以及个体到群体场景推演的研究路线。
- Focus
- Affective mind modeling, embodied cognition, long-term agent evolution, psychological digital twins, individual-to-group reasoning
- Application
- Digital humans, companion AI, counseling training, education, human-machine coexistence, organizational simulation
Mindmirror
Mindmirror is a face-to-face AI mirror for guided psychological assessment. It brings multimodal interaction, voice dialogue, posture and expression cues, and scale-informed profiling into a reflective assessment experience.Mindmirror 是面向面对面心理评估的 AI 心镜能力线，把多模态交互、语音对话、姿态与表情线索、量表启发的心理画像组织成一种可解释的反思式评估体验。
- Focus
- Multimodal assessment, reflective dialogue, psychological profiling, camera and voice interaction, report-level interpretation
- Application
- Schools, counseling rooms, organizational screening, healthy living environments, human-performance assessment
MindScore
MindScore extends assessment into organizational mind-body growth support. It connects non-contact multimodal sensing, long-term psychological archives, baseline comparison, trend awareness, and intervention workflow for institutions that need continuous care rather than one-off screening.MindScore 将评估能力延伸到组织场景中的心身成长守护，把非接触多模态感知、长期心理档案、基线对比、趋势感知和干预闭环连接起来，服务需要持续关怀而不是一次性筛查的机构场景。
- Focus
- Mind-body assessment, longitudinal profiles, baseline comparison, growth support, intervention coordination
- Application
- Education, enterprise health, public service, safety-critical teams, human-factor management
MindCare
MindCare translates affective computing into care systems. Representative outcomes include intelligent psychological monitoring, integrated mind-body assessment devices, and interactive intervention cabins that combine sensing, assessment, feedback, and intervention.围绕普适心理计算研发智能心理监护系统、智能身心评测一体机、智能心身干预座舱等成果，服务用户超过 20 万人。
- Focus
- Early screening, full-cycle mental-health management, interactive intervention, AI-assisted care
- Application
- Education, healthcare, elderly care, special populations, community mental-health services
Smart Cockpit and Human Performance
This direction is a representative deployment of the Ubiquitous Psychological Computing framework. It studies driver/passenger emotion, fatigue, stress, and cognitive load in mobility environments, then integrates perception, intervention, edge deployment, and cockpit interaction.这是普适心理计算框架的典型落地场景，面向驾驶员与乘员情绪、疲劳、压力和认知负荷建模，并把感知、干预、边缘部署与座舱交互整合起来。
- Focus
- Driver state monitoring, edge deployment, emotion-aware cockpit interaction, safety warning
- Application
- Automotive intelligence, public transportation, occupational health, human factors engineering
Child Development and Autism Support
Combining human-computer interaction, behavioral analysis, multimodal perception, and intelligent intervention, this project explores AI-assisted assessment and support systems for child development and autism-related scenarios.结合人机交互、行为信号、多模态感知和智能干预，探索面向儿童发展支持的辅助诊断、评估和陪伴式训练系统。
- Focus
- Behavioral signals, child-computer interaction, assistive diagnosis, individualized intervention
- Application
- Child development, special education, family support, clinical collaboration