Cases能力案例
Representative capability cases showing how MAC-Lab turns affective computing, psychological computing, and embodied emotional intelligence into deployable systems.展示 MAC-Lab 如何把情感计算、心理计算与具身情感智能转化为可部署系统的代表性能力案例。
From Research Route to Deployable Capability从研究路线到可交付能力
These cases help partners read MAC-Lab's capability map: what kinds of human-state problems we work on, how we organize sensing, modeling, evaluation, and support, and where a research idea can become a field-ready system.这些案例帮助合作伙伴理解 MAC-Lab 的能力图谱：我们面向哪些人状态问题，如何组织感知、建模、评价与支持，以及一个研究想法如何走向可在场景中验证的系统。
Representative Capability Cases代表性能力案例
Mindmirror / SoulMirror
Face-to-face multimodal psychological assessment for scenes that need evidence, interpretation, and a human-centered review process.面向需要证据、解释和人在回路复核的面对面多模态心理评估场景。
Mindmirror connects multimodal interaction, guided dialogue, scale-informed profiling, and report-level interpretation into a reflective assessment experience. The public value is clear: body, voice, expression, posture, and dialogue are not treated as isolated signals, but as converging evidence for a respectful assessment route.Mindmirror 把多模态交互、引导式对谈、量表启发画像和报告级解释组织成一种可复盘的评估体验。它公开展示的价值在于：身体、语音、表情、姿态和对话不是孤立信号，而是汇聚成一条尊重用户、可解释的评估路线。
- Capability shown展示能力
- multimodal assessment, camera and voice interaction, psychological profiling, reflective dialogue, report-level interpretation多模态评估、摄像头与语音交互、心理画像、反思式对谈、报告级解释
MindScore
Organizational mind-body growth support for continuous care, group insight, and intervention coordination.面向组织持续关怀、群体态势感知和干预协同的心身成长支持能力。
MindScore extends assessment into real organizational settings. It links long-term psychological archives, non-contact multimodal sensing, baseline comparison, trend awareness, and workflow coordination. The point is not to produce a single score, but to help institutions build a responsible route from observed evidence to growth support and early awareness.MindScore 将评估能力延伸到真实组织场景，把长期心理档案、非接触多模态感知、个体与群体基线、趋势感知和干预流程连接起来。它不是为了给出一个简单分数，而是帮助机构建立从证据观察到成长支持和早期感知的负责路线。
Guardian心身成长
守护
- Capability shown展示能力
- mind-body assessment, longitudinal profiles, baseline comparison, early awareness, intervention coordination身心评估、长期档案、基线对比、早期感知、干预协同
MindOS
A public-facing route for affective mind computing, psychological digital twins, and long-term embodied agents.面向情感心智计算、心理数字孪生和长期具身智能体的公开能力路线。
MindOS is the lab's affective mind-computing foundation. It can be understood as a backbone for agents that need state continuity, persona stability, embodied response, and individual-to-group scenario reasoning. The central idea is simple: emotional AI needs a mind-computing backbone, not only fluent language output.MindOS 是实验室的情感心智计算底座，可理解为支撑状态连续性、人格稳定性、具身回应和个体到群体场景推演的能力骨架。它的核心判断很清楚：情感 AI 需要稳定的心智计算骨架，而不只是流畅的语言输出。
- Capability shown展示能力
- affective mind modeling, embodied intelligence, psychological digital twins, long-term agent evolution, scenario reasoning情感心智建模、具身智能、心理数字孪生、长期智能体演化、场景推演
MindTalk
Simulation and training for counseling, supervision, group emotion, youth-support, and strategy exploration.面向咨询、督导、群体情绪、青少年支持和策略探索的推演训练能力。
MindTalk moves beyond single-session dialogue toward multi-domain psychological simulation. It helps show whether emotional AI can operate inside a scenario system: roles, pressure, group dynamics, intervention choices, and evaluation evidence all matter more than whether one answer merely sounds empathetic.MindTalk 不停留在单轮或单次对话，而是面向多域心理仿真与训练。它展示的是情感 AI 能否在场景系统中工作：角色、压力、群体动力、干预选择和评估证据，都比某一句回答是否显得“共情”更重要。
- Capability shown展示能力
- psychological simulation, counseling and training scenarios, group affect propagation, intervention strategy evaluation, research dashboards心理仿真、咨询与训练场景、群体情绪传播、干预策略评估、研究看板
MindPet
Embodied companion intelligence with affective memory, personality-aware interaction, and situated expression.融合情感记忆、人格化交互和场景化表达的具身陪伴智能。
MindPet explores how companion systems can stay coherent over time. Publicly, it shows that durable trust depends on continuity, restraint, relationship awareness, and multimodal expression; it is not just friendly dialogue wrapped in a cute interface.MindPet 探索长期陪伴系统如何保持连贯。公开表达的重点是：长期信任依赖连续性、克制感、关系感知和多模态表达，而不只是把友好对话包装成一个可爱的界面。
Agent陪伴
智能体 Expression表达 Safety安全
- Capability shown展示能力
- companion agents, affective memory, personality-aware interaction, group affect, embodied expression陪伴智能体、情感记忆、人格化交互、群体情绪、具身表达
MemOS-Mind
An affective memory layer for long-term AI interaction, service continuity, and psychologically aware response.面向长期 AI 交互、服务连续性和心理感知回应的情感记忆层。
MemOS-Mind makes the memory question sharper: a system should not only remember what happened; it should model what the event meant to the user and how the agent should respond through a stable persona and emotional stance. This supports companion AI, learning support, service assistants, and psychologically aware interaction.MemOS-Mind 把记忆问题说得更清楚：系统不应只记住“发生了什么”，还要理解这件事对用户意味着什么，以及智能体应如何以稳定人格和情感立场回应。这一能力支撑陪伴 AI、学习支持、服务助手和心理感知交互。
- Capability shown展示能力
- affective memory, user-state modeling, persona-consistent response, long-term interaction continuity情感记忆、用户状态建模、人格一致回应、长期交互连续性
From Capability Pattern to Partner Solution从能力案例到合作方案
Clarify the real problem behind the request找出需求背后的真实问题
A request for emotion recognition may actually be a need for safety warning, training evaluation, service improvement, mental-health support, or role-fit assessment.一个“情绪识别”需求，背后可能真正需要的是安全预警、训练评价、服务改进、心理支持或岗位适配评价。
Build a small route before exposing a large system先做小路线，再做大系统
The lab can help partners test signal feasibility, task definition, model route, user interface, and evaluation metrics before committing to full deployment.实验室可帮助合作方先验证信号可行性、任务定义、模型路线、用户界面和评价指标，再进入完整部署。
Evaluate in the field, not only in a benchmark在现场验证，而不只看基准数据
Field validation exposes sensor noise, missing data, privacy boundaries, user trust, edge constraints, and maintenance cost.现场验证会暴露传感噪声、数据缺失、隐私边界、用户信任、边缘约束和维护成本。
Leave reusable capability behind留下可复用能力
A good case should leave methods, modules, patents, evaluation protocols, student talent, and a clearer route for the next product iteration.好的案例应沉淀方法、模块、专利、评测流程、学生人才和下一轮产品迭代路线。
The valuable work is often in the middle layer.真正有价值的能力，往往在中间层。
Human-state systems succeed when data governance, model orchestration, evaluation, field adaptation, and long-term operation work together. MAC-Lab's cases are designed to make that middle layer visible and discussable.人状态系统能否真正落地，取决于数据治理、模型编排、评测、现场适配和长期运行能否协同。MAC-Lab 的案例设计，就是把这层关键能力变得可看见、可讨论、可合作。
Best for partners who need a route, not a slogan.最适合需要路线，而不是口号的伙伴。
MAC-Lab is a good fit when a partner has a real scenario, sensitive human-state questions, and a need for research credibility, engineering realism, and deployable solutions to mature together.当合作方有真实场景、敏感的人状态问题，并希望科学可信度、工程现实性和可落地方案共同成熟时，MAC-Lab 最适合参与。