In just twelve short hours, Pejman Ghorbanzade assembled an array of wearable sensors to create “Sloth,” a project that won first prize at IoTHackDay 2016 in Minneapolis. This sensor-based activity recognition system blends wearable computing with Internet of Things using Digi XBee® modules to create a mesh network of small-sized embedded sensor platforms that can identify daily activities such as cooking, walking, sitting down, etc.
At IoTFuse, the largest IoT conference in the midwest, Ghorbanzade presented his project and explained that by attaching a few accelerometers to different body parts, he can collect human physical motions and transmit them to a central node where we can interpret the type of activity being performed and transmit it to the cloud where it can be accessed by privileged users. This type of wearable tech could be a game changer for healthcare monitoring systems and assisted living applications.
We asked Ghorbanzade what inspired the creation of Sloth, why he chose Digi XBee, and how he overcame design challenges and criticism:
When did you first start working with Digi XBee and why did you start?
My first experience with Digi XBee was in November 2012 when I started my research in wireless sensor networks during my undergraduate studies using a Waspmote Starter Kit. The kit included a sensor platform and a communication shield from Libelium and two XBee modules.
I clearly remember my excitement when I managed to wirelessly transmit power readings from the sensor platform to my computer using XBee modules. Since I had only one board, the applications I could work on were fairly limited, but soon I created a simple system for fall detection of the elderly which perhaps sparked the initial idea of developing Sloth.
Who or what inspired you to create Sloth?
The idea of building a wearable activity recognition system started to develop when I was plotting accelerometer data in tri-dimensional space. Initially, I was not aware of the potential of such system in healthcare monitoring and assisted-living and I was not building the system with a specific application in mind.
However, as the system started to take shape, my colleagues started to ask what it is and what applications it is used for. This helped me gradually realize how such system might be useful for improving the quality of life of the elderly. To think that my system could have such impact was itself a great motivation through the design iteration and development process. It also motivated me to better understand the requirements of personal healthcare monitoring systems and shape the product accordingly.
All of this inspired me to build Sloth; a real-time system that allows physicians and authorized users (e.g. immediate family members) to monitor activities of daily living and get notified when certain conditions are met. I firmly believe that such system could be widely adopted in the near future, considering that our societies are aging and the cost of traditional healthcare monitoring solutions such as home nursing is becoming increasingly prohibitive.
Why did you choose to use Digi XBee for this project?
Sensor-based activity recognition is not a new concept and has been well-researched over the past few years. There are already commercial products in the market and the research community is actively publishing new works on different techniques to detect activities of daily living. Consequently, I soon realized that the only way to make a meaningful contribution in development of wearable systems is to identify and address problems that are less investigated and yet are of great practical value.
One of these problems is energy-efficiency. Wearable systems are expected to have long operational lifetime despite limited power resources. This is especially challenging for wearable activity recognition systems since detecting activities of daily living requires collecting motion data at high sampling rates. At the same time, these systems are expected to function in real-time so there needs to be a continuous flow of wireless transmissions among sensor nodes.
To satisfy these requirements, I had to choose a wireless module that is easy to integrate with my prototype boards, reliably transmits data packets, supports various mesh network topologies and, more importantly, consumes the least amount of power in the long run. After careful consideration of different wireless modules and comparing their results, I chose Digi XBee for my project.
What tools do you find indispensable for accomplishing a project with Digi XBee?
Digi XBee modules are easy to integrate with almost any off-the-shelf board which makes them straightforward to set up and easy to communicate using them. This also removes the need for extra tools and modules. For configuring wireless mesh networks, Digi’s XCTU software makes it easy to define the network topology and inter-network communication policies.
What do you find are your biggest stumbling blocks and what are the best ways you’ve found to overcome them?
My biggest challenge was to design a distributed algorithm to minimize network communication rate while maintaining real-time functionality of the system. This required development of a specialized algorithm that can partially process accelerometer data that are collected at very high sampling rates on individual sensor platforms with fairly limited computational capacity.
This made me thoroughly investigate various machine learning algorithms to better understand their characteristics. After many months of research, I finally developed the algorithm that is used in Sloth today and is specialized to minimize power consumption of the system.
What’s your best advice for handling criticism?
I think for any project idea, it is crucial to hear different opinions and spend time thinking about them. I have tried to explain my project’s idea to anyone that is interested in hope that I can follow their line of thinking to identify the missing links in the story. It is also essential to always admit that your product is far from perfect and spend more time investigating its deficiencies than cherishing its advantages.