Home
|
|
Frequently Asked Questions (FAQ)
What are wockets?
Wockets are miniature sensors we are developing as part of this open source hardware and software project. Our goal is that wockets will do one thing really well: get 3-axis accelerometer data into common mobile phones at a sufficiently high sampling rate to enable continuous real-time activity recognition ... but in a form factor that can be worn 24 hours per day. We want them to be affordable, easy to obtain and/or build, robust, and ergonomic. Multiple wockets should work with the same phone, allowing application developers flexibility.
Which phones work with wockets?
Wockets will send data via Bluetooth to phones, therefore, in theory, it should be possible to get them to work with nearly any current phones. In practice, the type of Bluetooth stack used by the phone as well as the way the phone does memory management may impact how viable the phone model is. We hope, however, with community participation, that we can eventually provide code that makes it easy to use wockets on many different types of phone operating systems, and many different devices within each OS. As a starting point, we are doing software development and testing on Windows Mobile phones. We welcome participation from experts in the other common operating systems. Eventually, once we have fully developed and tested the system using Windows Mobile phones, we plan to port the wockets software to at least one other OS.
With what do you need help?
Many things, including... (1) critique of overall system design, (2) design and optimization of the hardware, (3) design and optimization of the strategy for extended Bluetooth battery life, (4) critical analysis of the hardware to eke out cost savings, (5) finding device manufacturers, (6) developing cool example applications that use accelerometer data on the phone, (7) optimizing code on the phones so that it runs efficiently and robustly, (8) identifying other emerging devices related to wockets, (9) porting code to different phone models and phone operating systems, (10) methodical testing and characterization of the various components of the system (e.g. wireless properties), (11) developing strategies for ensuring data security, (12) developing or porting algorithms for real-time activity detection to the mobile phones, and more.
Why are you developing software using Windows Mobile?
The MIT research group that started this project has a substantial amount of experience developing software for Windows Mobile devices. We have found, as have some of our colleagues at other major research universities, that these devices, despite some problems, do allow for very rapid prototyping of complex software. The OS puts few limits on what one can do, and the development tools are good (e.g., allowing the same code to run on the PC and the phone, allowing very nice on-device runtime debugging). Is it perfect? No. Do these devices have the largest market share? No. However, the learning curve is definitely less steep than on some other phones, and the developer has far more flexibility in some cases.
Are there plans to develop software for other phone platforms?
Yes. We plan to port the software to one other phone platform. If we had to pick today, that platform would be Android, because it appears to be quite good for advanced prototyping. However, if Android does not catch on within the next two years, we would select another OS.
Why not develop for the iPhone?
There are many things about the iPhone that are impressive. Unfortunately, there are two aspects of the phone that are problematic for this project. (1) It is not possible to run a background process. This means the phones can not perform continuous activity recognition 24/7 and save and process Wockets data. (2) The Bluetooth is setup only for wireless headset modes ... unless Apple decides to provide API support. If iPhone gurus think they can hack around these issues, though, please join us!
How hard is it to use wockets?
Our goal as the project goes forward is to make it very easy to incorporate wockets (or output generated from the activity detection software) directly in your own applications or (if you are a researcher) experiments. Right now building and using them requires a fair amount of expertise, but that will change as we go forward.
Where can I get wockets?
Eventually, we will provide all the information you need so that you can easily order wockets in small or large quantities from a hardware assembly company. It won't quite be "one click," but we'll do our best to simplify the process. We aren't there yet, though, because we are still developing the hardware. For current versions, you will need electrical engineering expertise. In that case, all the information you need to make them or have them made for you can be found on the wiki.
Can I build my own wockets?
Wockets will definitely use surface mount components. The will therefore be difficult, but not impossible, for someone to solder by hand, with the right tools. We will make sure all the information needed to build them by hand is online, as well as tips on how to do so. However, we expect that the easiest and least expensive way to obtain them will simply be to order a few already constructed.
How much do wockets cost?
We don't know yet, but the goal is to make the software free and the hardware as inexpensive as it can be. For real-time streaming of data to the phone, wockets should be the least expensive option available. One area where you could get involved is in helping us find ways to keep the cost of manufacturing Wockets down.
How do I contribute code to the wockets project?
Our code base is currently on Google Code but not quite organized enough yet to make it easy to get started. Stay tuned as we work on that. We'll use SVN as in most open source software projects. We are interested in optimizing the code that handles sensor connections, battery life, and data saving/processing, as well as the code that uses pattern recognition for real-time activity training and recognition.
Why should I contribute to the wockets project?
Despite many devices and software out there, it is still not easy and affordable to create technology that uses accelerometry in novel ways on mobile phones. It is even more tricky to write robust software to detect physical activity. By contributing to this project, you will help to create the tools needed so you and others can create some truly innovative mobile applications, either for health, communication, or fun. The more people who use and contribute to the development of the devices, the more affordable they will be. If the devices begin to be used in large medical studies, you can feel good knowing that you played a role in helping to advance health research that may ultimately lead to new insights into how to help people stay healthy throughout life.
Why not just use one of the many sensors that research groups and companies have already developed?
There are certainly many wireless accelerometers already out there. However, to our knowledge, none meet the goals of this project for sensors that are sufficiently small to wear 24/7, allow data to get into common mobile phones without special/awkward connections, allow multiple data streams to be streamed to the phone in real time, work robustly and don't require any manual operation for pairing/rebooting, and can be obtained at the lowest possible cost. The other options tend to be too big or have very limited battery life. They also tend to be quite expensive or difficult to obtain in small and large quantities. Many of the smallest devices do not stream high sampling rate data in real time to the phone. Others use wireless protocols that only work with one or two phone models or that require special add-ons to the phones themselves. There is an important niche for both hobbyists and researchers that no currently available hardware seems to meet.
Why Bluetooth? Why not a low-power wireless protocol?
We have debated about this. At least for now, we decided, after doing a lot of testing, that using several strategies we can eke out enough performance from Bluetooth to meet a goal of 24h performance on one charge with a sensor form factor that meets usability constraints. We still need to prove this (hopefully with your help!), but this design option seems less problematic than the other option of using a low power protocol with an intermediate device that collects data and then transmits to the phone (via Bluetooth). See the wiki for more information and to contribute your thoughts.
Nearly all future mobile phones will have accelerometers built in, so why are wockets necessary?
There has been some nice research recently showing reasonably good activity recognition performance using only the sensors that are already built into the mobile phones. For those people who carry mobile phones attached to a belt or in their pockets, it will definitely be possible for the phones to detect some activities (e.g., sedentary vs. walking). However, despite some of the results that have been reported, research at MIT suggests that adding 1-2 additional wireless accelerometers may dramatically increase robustness as well as the number and variety of activities and activity intensities that might be automatically detected by the phone. Moreover, not everyone will be willing to carry a phone in an "optimal" way for activity detection. It may be more convenient to wear a specially-designed sensor underneath clothing, especially if the phone can do many interesting things for someone if they do. All that said, the wockets phone software will permit the use of wockets only, the phone's internal accelerometer only, or both.
Why in the world would someone wear Wockets 24/7?
Well, if they are well-designed, the answer will be, "why not?" They will be so small and comfortable and convenient that the benefits will far outweigh the risks. Benefits might be (1) feeling good from participating in a research project that helps improve health, (2) benefiting from software on the phone that might help with weight maintenance by tracking physical activity, (3) using software that helps with time management in innovative ways using the automatically-detected activities, (4) having fun playing games that use body movement, (5) experiencing fewer inconvenient interruptions from the phone because it automatically detects inappropriate times to interrupt, (6) sharing information about one's own activity with friends and family in new types of social networking applications, and more. One of the goals of this project will be develop a system that is so convenient and useful that people miss it when they don't have it and therefore make efforts to wear the sensors every day. We'd like to hear from you about this. Viable?
Why is this project funded, in part, by the National Institutes of Health?
Medical researchers are increasingly interested in studying gene-environment interactions. They want to know how genetic predisposition as well as lifestyle choices impact short and long term health. Unfortunately, tools for studying behavior of people in their natural environments are limited. Mobile phones that could reliably detect the type, duration, intensity, and location of physical activity could create exciting new opportunities for researcher to better understand the relationships between physical activity, sedentary behavior, and health and wellness. The wockets system would be such a tool and could be widely deployed in future studies if the system is sufficiently robust and affordable and if subjects in experiments are willing and able to wear the sensors for weeks, months, and even years.
How do wockets relate to MITes?
You may have heard of MITes (MIT Environmental Sensors). Those are an earlier version of wockets developed at MIT by the House_n research group. The MITes use Nordic radios. They are still be used for research by House_n (to test Wockets, among other things) and the designs are available to you if interested, but we are no longer making new ones.
How do I add my own sensors?
The focus of this project will be on real-time accelerometry and making the devices as small and as affordable as possible. That said, it will be possible for you to take the wockets designs and tweak them in whatever way you choose create new sensor devices. Our goal is not, however, to create a general sensing platform as some other projects have done (e.g., Arduino), but instead to create a hardware and software system highly optimized for a particular task.
|