Background research

Osman Ullah
Rahul Nair

The aim of our project (Counter Activity) is to track user interaction on a countertop using RFID. We chose the cooking application because similar work is already being done in the Aware Home [2][3][4](Cooks collage).

Initial plan
Our initial idea was to tag everything in the environment and then have readers mounted underneath the counter facing upwards and under the cabinets facing downwards. By knowing the knowing the area of coverage for each antenna, we could track tagged items as they were moved around the countertop. This information could then be used to find out what the user is doing on the countertop. Current work is moving towards using computer vision[6] to track the user information. While this provides the advantage of no extensive wiring requirements, we would like to try using RFID because we believe that RFID will give us better reliability than computer vision in it's current state. RFID also has a proven track record with work such as Smart Playing cards[5] and Listen Reader[7].

The two primary concerns with the RFID technology were the read range and anti-collision. Typical RFID systems do not have a read range of more than a few inches. This would pose a problem because if the read range was not high enough, it would create 'dead zone' in between the top and bottom readers. In this dead zone we would not be able to know what the user is doing. Anti-collision technology is important because there will be multiple items in the range of readers at all times.

We came across a system from Checkpoint Systems which is the new 13.56 Mhz technology. These tags have anti-collision capabilities and the readers have a range of 12-24 inches for the general purpose model. We could also turn down the power of the readers if the range was not short enough to give us the granularity we needed. However, these systems are expensive, and the tags are proprietary.

The other system we are evaluating is the Texas Instruments Tag-It line of products. These also used the 13.56 Mhz standard, and they are ISO/IEC 15693 compliant allowing use of tags from multiple manufacturers. In addition they have a relatively inexpensive starter kit. The range of these readers, however, is only 18 cm. This will be lessened by environmental variables such as metal and obstacles. The Tag-It system has also been used successfully in situtions where a single reader has to read multiple tags and handle collisions[9].

Additional constraints
After doing further research, however, we realized two more problems-cost and interference. The readers are not cheap and having an array of multiple readers underneath the counter and cabinets would be very expensive. In addition, it is very likely there would be issues with interference betwen readers. The readers are creating magnetic fields to power the tags. If these magnetic fields interfere with one another it would lead to unpredictable results.

The best solution we have come up with so far is to try to use a single reader mounted on the user. We can separate the reader and antenna, create a custom antenna and mount this antenna on one wrist of the user. A similar idea has been tried by Gellersen and Merz at the University of Karlsruhe in Germany. We can then tag the counter top and all the items in the kitchen. This method will not give us the location of the items on the countertop, but we can still know what items the user is interacting with at any given time. This is really the information we need. The only drawback of this system is the reader is not wireless and has to be connected to a computer. However, as you will read later we think we have found a solution to this problem.

The need for movement tracking
While we can now figure out what items the user is picking up we do not quite know what the user is doing with them. We don't know if the user has poured the item into the bowl, or mixed items together in a cup, or is stirring things. The best solution we think is to mount an accelerometer on the user's wrist. Since the hand will be moving in 3d space we need a 3-axis accelerometer. We considered the motes, but they are only single axis. Phidgets were also looked at but they are only 2 axis. We finally came across the HP Badge4 prototypes. Amongst many other sensors, a 3 axis accelerometer is included on the board. The board is capable of wireless connectivity with a PCMCIA 802.11b adapter and also has an RS232 interface. If we can use these we will not only have the 3-axis accelerometer, but will also have a wireless connection point for the RFID reader.

Current system plan
Mount an RFID antenna in the users palm or wrist (maybe using a glove or directly strap it to the palm) and connect it to the RFID reader using a cable. While the reader itself comes with an AC adapter we plan to wire it to a 9V battery and make the reader a self contained unit. The HP Badge 4 will also be strapped to the users wrist and will have a PCMCIA card to provide wireless connectivity. We plan to interface the RFID reader to the Badge 4 and use it to transmit the RFID reader data to a different computer which will do the actual detection of the tasks on the counter.

References
Cooks collage
Equator project
Checkpoint Systems
Texas Instruments 6000 series reader
HP Badge4 System
[1]Gellersen and Merz, Enabling Implicit Human Computer Interaction A Wearable RFID-Tag Reader
[2]Tran, Q., Mynatt, E. (2003). "What Was I Cooking? Towards Deja Vu Displays of Everyday Memory."
[3]Tran, Q., Mynatt, E. (2002). "Cook's collage: Two exploratory designs."
[4]Mynatt, Essa, Rogers "Increasing the opportunities for aging in place"
[5]Romer, Domnitcheva "Smart Playing Cards: A Ubiquitous Computing Game"
[6]Fails, Olsen "Light widgets: interacting in every-day spaces"
[7]Back, Cohen, Gold, Harrison, Minneman "Listen reader: an electronically augmented paper-based book"
[8]Want, Fishkin, Gujar, Harrison "Bridging physical and virtual worlds with electronic tags"
[9]Back, Cohen "Page detection using embedded tags"

Link to this Page

Counter Activity last edited on 12 February 2004 at 1:18 am by c-24-98-172-105.atl.client2.attbi.com