同济大学设计创意学院 · 博士在读
PHD CANDIDATE · COLLEGE OF DESIGN AND INNOVATION, TONGJI UNIVERSITY

陈乐恒. LEHENG CHEN / LOKHANG CHAN

我做人机交互研究，把拓展现实与可穿戴、触觉、人工智能带进康复训练 —— 让它更精准、更智能，顺便变得有点意思。

I build XR, wearable and AI systems for cognitive and motor rehabilitation — making it more precise, more intelligent, and along the way a little more worth doing.

XR
Healthcare
Wearable
Haptics
Artificial Intelligence

01 — 关于 / ABOUT

你好，
我是乐恒。Hello, I'm Leheng.

ORCID: 0000-0002-8297-6925
LOCATION: 上海 Shanghai · 阜新路281号
EMAIL: lokhang@tongji.edu.cn

土木出身，提桶跑路做交互设计 —— 从物理搬砖到赛博搬砖，中间只差了亿点点对打灰的恐惧。现于同济大学设计创意学院攻读先进技术与设计方向博士，研究人机交互中的拓展现实与人工智能如何让认知与运动康复更精准、更智能。

I started in civil engineering and walked off the site into interaction design — from moving actual bricks to moving cyber ones, the gap being roughly one lifetime's worth of dread about pouring concrete. I'm now a PhD candidate in Advanced Technology and Design at Tongji University D&I, working on how extended reality and artificial intelligence in HCI can make cognitive and motor rehabilitation more precise and more intelligent.

篇论文
PAPERS

第一作者
FIRST AUTHOR

论文奖项
PAPER AWARDS

非学术特长 / OFF-DUTY

🀄 立直麻将爱好者（雀魂豪1，主营放铳）Riichi mahjong — Majsoul Jade 1, mostly dealing in
📷 人像摄影，尼康党Portrait photography, shot on Nikon
🎮 深度电子杨伟Deeply committed video-game enjoyer

02 — 研究 / RESEARCH

一个主题：健康ONE THEME — HEALTHCARE

我关注一个具体问题：康复训练枯燥、反馈迟钝、难以坚持。工作沿三条线索展开 —— 拓展现实负责造出值得投入的情境，传感器负责把场景中身体与认知的真实状态读回来，人工智能把传感数据转化成更个性化的反馈与交互。

One problem drives my work: rehabilitation training is dull, its feedback is slow, and people quit. Three threads run through everything I build — extended reality creates a context worth engaging with, sensors read the real state of body and mind back out of that context, and artificial intelligence turns that sensor data into feedback and interaction tuned to the person.

三项技术手段 / THREE MEANS

01 / 手段

拓展现实EXTENDED REALITY

康复训练最大的敌人是无聊。XR 让研究者能造出可控、可重复、可量化的情境 —— 同一个任务可以精确复现无数次，难度可以连续调节，每一次表现都留下数据。它也能改写使用者所看到与感受到的东西：视觉可以被引导，本体感受可以被欺骗，一个动作的反馈可以被放大到刚好足够察觉。

The biggest enemy of rehabilitation is boredom. XR lets a researcher build contexts that are controllable, repeatable and measurable — the same task reproduced exactly, difficulty tuned continuously, every attempt leaving data behind. It also rewrites what a user sees and feels: vision can be guided, proprioception can be fooled, the feedback on a movement amplified to just past the threshold of noticing.

AR / VR / MR
Visual Illusion
Spatial Interaction
Interaction Channels
Presence & Embodiment

02 / 手段

可穿戴传感WEARABLE SENSING

要判断一次训练是否有效，先得知道身体真正在做什么。可穿戴设备可以贴着皮肤持续采集人体数据，不挑场地，也不会让人觉得自己正在被观察。反过来，可穿戴也可以是输出 —— 把反馈直接渲染到皮肤上，不去挤占已经很拥挤的视觉与听觉。

To know whether a session worked, you first have to know what the body actually did. Worn against the skin, a wearable collects data from the body continuously — no dedicated space, and no feeling of being watched. It works the other way round too: a wearable can be an output, rendering feedback straight onto the skin instead of crowding the already busy visual and auditory channels.

Textile Sensing
Wearable Prototyping
Motion Estimation
Physiological Signals
Haptics

03 / 手段

人工智能ARTIFICIAL INTELLIGENCE

传感器给出的是一串数字，使用者需要的是一句「再抬高一点」。人工智能负责中间这段转换：从连续信号里认出动作、判断质量、预测一个人什么时候快要放弃，再决定此刻该提示什么、该沉默多久。它也让训练方案不必是一张固定的表格 —— 难度、节奏、鼓励的语气都可以跟着人变，而不是让人去适应系统。目标不是让系统显得聪明，而是让每一次反馈都恰好够用，不喧宾夺主。

A sensor produces a stream of numbers; what a person needs is “lift it a little higher.” AI does the conversion in between: recognising movements in a continuous signal, judging their quality, predicting when someone is about to give up, then deciding what to say now and how long to stay quiet. It also frees a training plan from being a fixed table — difficulty, pacing and tone can follow the person rather than the person adapting to the system. The aim is not a system that looks clever, but feedback that is exactly enough and never takes over.