IO Armrest 浴室跌倒检测扶手

According to WHO, 44% of the elderly have ever fallen in the bathroom, with over 80% of Empty-nesters can’t get first-aid in time. To help them achieve more private space, it took hundreds of days to come up with this revolutionary patent solution. Without using cameras or wearable devices, IO Armrest use ToF Sensors to monitor the horizontal space and make judgments through Deep-Learn algorithm. If falls happen, it auto contact emergency service. IO Armrest is the first low-cost and privacy-free bathroom fall detection product. Developed by interdisciplinary design & engineering students, and provides a universal solution for aging problem.

根据世界卫生组织的数据，有44％的老年人曾经跌倒过洗手间，其中80％以上的空巢老人得不到及时的急救。为了帮助他们获得更多的私人空间，这种创新的专利解决方案花了数百天的时间。 IO Armrest在不使用相机或可穿戴设备的情况下，使用ToF传感器来监视水平空间并通过Deep Learning 算法做出判断。一旦发生跌倒，它将自动联系紧急服务。 IO Armrest是世界上第一个低成本且无隐私的浴室跌倒检测产品。由跨学科设计和工程专业的学生合作开发，并为人口老龄化社会提供了一种具备潜力的解决方案。

HOW IT IS DIFFERENT 它有什么不同？

• Compared to a privacy-infringing camera fall recognition solution, we have developed the first new fall detection solution and algorithm on the market that do not violate privacy and has low cost and low power consumption.

• Empower the elderly-use smart homes to assist the independence of the elderly, return the right to use technology to the elderly, so that the elderly also have dignity.

• Respect for privacy-also enjoy proper private space and the right to take a bath alone.

• 相比于侵犯隐私的摄像头跌倒识别方案，我们研发构建了市面上第一种不侵犯隐私且低成本 低功耗的全新跌倒检测方案与算法。

• 为老人赋能—用智能家居辅助老人独立，将使用科技的权利归还老人，让年老也有尊严。

• 尊重隐私—同样享有适当的私人空间和独自洗浴的权利。

Facts About Empty Nesters 空巢老人现况

My grandpa accidentally fell while taking a bath alone before. Fortunately, my mom discovered and offered timely assistance. But investigation revealed that there are over 130 million elderly people living alone in China, not everyone is as fortunate as my grandpa. In fact, more than 80% of them were hurt worse because of untimely assistance. We want to help them so we use ToF to make convenient and economical armrest detect falls in the slippery bathrooms. As an aging society comes, we want to empower the elderly, using technology to help them become more independent and retain dignity.

几年前姥爷独自洗澡时意外跌倒，所幸家人第一时间发现，姥爷得到及时的援助。因此时⻓ 出现在新闻中的独居老人意外跌倒今日了我的注意。随着深入调查发现，我国有超1.3亿独居老人，他们在家中意外跌倒时超80%因无法及时发现得到救助而造成更严重的伤害，其中绝 大部份的跌倒发生在湿滑阴暗又隔音的浴室。现有的跌倒检测方案都借助摄像头或可穿戴设备，难以适用于浴室且普遍价格不菲。在全球老龄化日益加剧的今天，我们想为老人赋能， 运用科技手段帮助他们更加自主独立，保留尊严，而非剥夺隐私和生活。

How Technology Might Help? 科技如何帮助？

I bring this topic to Microsoft Asia R&D Group And discuss with top CS students from all the top university in China to explore the possibility of Design + Technology , and here are what we came out:

我将这一话题带去了2019年微软夏令营，与国内各知名高校的程序开发者们共同探讨设计+科技相结合的可能性，并得出两个重要观点：

IO Armrest detects fall through our patented solution. We changed tradition way of detecting and only sensed the distance between each point of the armrest to the person so that fall detection can be completed without cameras or wearable devices. IO Armrest emits multiple IR beams in a horizontal space at specific frequency during operation. The space distance is calculated inversely according to the flight time of light into a dynamic chart for the Machine Learning algorithm to detect falls. The cloud server determines the user's behavior mode according to the curve change, and distinguishes between bending, leaving, and falling. Users only need to fix the handrail at a height of 1M from the ground next to the shower, and will automatically monitor when taking a bath. Once the system judges a fall, it will send warning signals to the pre-set community carers or relatives immediately. IO Armrest empowers elderly people living alone to enjoy private space.

IO Armrest 通过我们耗时2年独立研发的专利方案判断跌倒。传统的图像识别跌倒判断方案不 仅侵犯隐私，并且功耗巨大位置受限。我们转变思路，仅感知1米高度处扶手各点与人的距 离，从而无需摄像头或可穿戴设备即可完成跌倒检测。IO Armrest 在工作时按特定频率在水 平空间平行发射多束红外光，根据光线⻜行时间反算空间距离，组成水平空间动态距离图并 交由机器学习算法判断跌倒。云端服务器可以根据曲线变化情况判断用户的大致行为模式， 并对弯腰、离开与跌倒三种状态进行区分。用户只需将扶手用强力胶固定在花洒旁距离地面 1M高度处，安装电池并联入Wi-Fi 网络，无需任何操作在洗澡时即会自动开始持续监测。一 旦系统判断发生意外跌倒无法起身，将会第一时间发送求助信息给预先设定的社区护工或亲 属，并电话呼叫接通预设的紧急联系人。IO Armrest 使独居老人也可享有适当的私人空间和 独自洗浴的权利。

We endow traditional homes with intelligent functions, reduce the learning cost of old people for technology products. The challenge and difficulty of the bathroom scene are: wet and difficult to transform. For the sake of privacy protection and low power consumption, we convert the change of human posture into a fluctuating curve. To this end, we formed an interdisciplinary team and began to independently construct an "artificial intelligence enhanced sensor" for privacy protected monitoring of falls. We noticed that the bathroom handrails provide a suitable position, so a test prototype was constructed on the handrail. With the help of 4 TOF ranging sensors and ESP8266 IoT chips, the cost of hardware is controlled within €10. Then we designed the initial appearance and structure of the IO Armrest smart armrest. After data analysis and model training on Azure cloud server, we made the circuit board of the product and made it with a complete appearance, functional and waterproof structural user test prototype. It took nearly two years and tens of thousands of funds to make it happen. After six iterations, we achieved 95% detection accuracy in several rounds of user tests, which met our psychological expectations.

我们赋予传统家居智能化功能，降低老人对科技产品的学习成本并提高其可接受度。浴室这 一场景的挑战和难点在于:潮湿、且难以改造。我们为了方案隐私保护且低功耗将人体姿态 的改变转换为一条波动的曲线录入电脑机器学习。为此我们组成了跨学科研发团队，开始独 立架构用于无感监测跌倒的“人工智能增强形传感器”。我们注意到老人本应需要的卫浴扶手恰 巧提供了合适的高度位置与内部空间，且减少跌倒概率与跌倒检测可以构成功能拓展，因此 在扶手上构架了测试原型。借助4枚TOF测距传感器以及ESP8266系列物联网芯片，硬件部分 成本被控制在10欧以内。随后我们设计了IO Armrest 智能扶手产品的初步外观与结构，在 Azure 云服务器上进行数据分析与模型训练后，嵌入式工程师制作了产品的电路板，并将其制 作成具有完整外观、功能与防水结构的用户测试原型。前后耗时近两年时间，花费数万经 费，经历前后6次迭代，我们在几轮用户测试中达到了95%的检测准确度，符合我们对其的心 理预期。

We are currently in the 7th product iteration to estimate the cost of mass production, and optimize the ToF monitoring algorithm to make it more rapid and accurate to improve monitoring accuracy and reduce latency. We are also conducting long-term use tests that last for half a year to evaluate water resistance and long-term reliability. Several major communities and smart home companies have contacted us to seek further cooperation and are willing to assist us improving the product.

我们目前正在第7次产品迭代，以估算量产成本，进一步优ToF监测算法，使其提升监测准确 度并减少延时反应更加迅捷。我们也正对其进行持续半年的⻓期使用测试，用于评估防水性 与⻓期可靠性，几家主要社区与智能家居企业已经与我们联系以寻求进一步合作，并愿意辅 助我们进一步完善该产品。

We have one invention patent "fall detection method, device and storage medium" and obtained several software copyrights to protect the core algorithm results. The project was supported by Hunan University and Microsoft Azure and was shared with professions from Microsoft Garage at the Microsoft Summer Camp in Beijing.
我们已经针对我们的研发成果申请了1项发明专利《跌倒检测方法、装置及储存介质》CN 111012359 ，并获得数项软件著作权用于保护核心算法成果。该项目获得来自湖南大学及微 软Azure云的经费支持辅助，并曾在北京微软夏令营上与来自微软创新⻋库的老师进行分享和交流讨论。

本项目为4人项目，扮演全链路设计师及项目负责人角色

负责：

1. 提出前期研究目标——以温和科技形式，在不侵犯隐私的基础上为老人浴室跌倒问题提出解决方案

2. AI及传感器技术解决方案研究——与程序和嵌入式工程师合作，提出转变视角数据降维的概念，摒弃传统摄像头视角方案，重新构建基于扶手的传感检测方案，使其同时满足成本低廉、不侵犯隐私、检测识别率高等特性，该方案已获发明专利。

3. 设计产品ID造型、CMF方案及结构装配设计，并与手板工厂合作完成样机制作。

4. 负责建模渲染及产品交互设计，并通过用户测试和可用性评估，前后共53版本迭代造型结构，26次迭代交互效果。

5. 以设计师及研究者身份，与程序及开发人员共同合作完成研究及项目开发。

6. 以项目负责人身份，协调队友跨学科高效交流、协同创新。