An embedded MEMS sensor might be lots of fun to play with on your first foray into the embedded world–why not deploy a whole network of them? Alas, the problem with communicating with a series of identical sensors becomes increasingly complicated as we start needing to handle the details of signal integrity and the communication protocols to handle all that data. Fortunately, [Artem], [Hsin-Liu], and [Joseph] at MIT Media Labs have made sensor deployment as easy as unraveling a strip of tape from your toolkit. They’ve developed SensorTape, an unrollable, deployable network of interconnected IMU and proximity sensors packaged in a familiar form factor of a roll of masking tape.
Possibly the most interesting technical challenge in a string of connected sensor nodes is picking a protocol that will deliver appreciable data rates with low latency. For that task the folks at MIT Media labs picked a combination of I²C and peer-to-peer serial. I²C accomodates the majority of transmissions from master to tape-node slave, but addresses are assigned dynamically over serial via inter-microcontroller communication. The net effect is a fast transfer rate of 100 KHz via I²C with a protocol initialization sequence that accommodates chains of various lengths–up to 128 units long! The full details behind the protocol are in their paper [PDF].
With a system as reconfigurable as SensorTape, new possibilities unfold with a solid framework for deploying sensors and aggregating the data. Have a look at their video after the break to get a sense of some of the use-cases that they’ve uncovered. Beyond their discoveries, there are certainly plenty others. What happens when we spin them up in the dryer, lay them under our car or on the ceiling? These were questions we may never have dreamed up because the tools just didn’t exist! Our props are out to SensorTape for giving us a tool to explore a world of sensor arrays without having to trip over ourselves in the implementation details.
via [CreativeApplications]
I can see this in all sorts of crowd/queuing control applications.
I have a project this would be ideal for now. But, not available yet. Oh well.
And now the wait for it to come to market. This could either be already, given that the paper is a couple months old and there are thousands of manufacturers on several continents that would take something like this and run with it, or forever, if the cost can’t be brought low enough for anticipated volume.
An excuse to cover a robot arm in sticky tape.
Yeahhh, you know, this tape is not there because I used cheap RC servos for my robot arm and it keeps falling apart, it’s there because, umm, uhm, sensors! you dig?
Usually seems to hold better than hot glue, not as messy as gorilla glue and safer for eyes and fingers than super glue if you’ve run out of solvent. [probably looks a little neater too/authentic home made hackernel look.] haha
^damn, auto correct and lack of proper delete facility of wordpress.
Nice.
I don’t want to be cynical, but is this any different than a really long Kapton circuit you can roll up?
Physically not really different, IMHO you are right there. The difference is how the ICs organise themselves regarding the I2C bus, which makes these tapes easy to interface and integrate to a µC.
I do like the right angle cut lines that you can use to solder right angles into the tape… Would be nice on the LED tape as well.
OK now that is kinda stupid awesome and I want it :D
If I only had the Money to play with all the things that are coming out now.
And I wish I was my Kids age………….
Oh well Have to play with what I have now…
This is so cool.
Flexible and QFN. that looks like a nice marriage…
That’s really cool, I especially like the cut lines and the thought that clearly went into making them able to be joined up in different ways and at different angles, and how 3 pads on one side + 2 on the other guarantees not messing up which pad goes where when connecting segments.
seems incredibly cool to play with ! However I wonder about the price
It’s sell like hotcakes! …….What?! Only 12 feet exist? $10000 a foot then!
from the pdf:
Cost
Cost is an important consideration, since SensorTape involves
a large number of electronic components, and has more value
at a large scale. The price of electronics is the major cost:
one node with the IMU is about $5 (US Dollars). Fabrication
of the flex circuit for one node is about $1.6. So, the price
is about $100 per meter. The price of electronic components
was about half of the cost of the large electronics distributors
(e.g. DigiKey), because we ordered in quantity of 100 from
a China-based supplier, found on http://www.alibaba.com. On the
downside, we spent a lot of time communicating with the suppliers,
as there were language barriers. The proximity sensor
we are using is novel, and therefore costs almost as much as
all other parts ($4), increasing the price to $200 per meter. It
is possible to bring the cost down by integrating the computation
and sensing functions into a single chip. Also, it may
make economical sense to have different tapes, each with just
proximity or IMU, as users might need only one sensor.
I’m surprised that I2C can run 128 nodes, seeing how it only uses pull-up resistors to drive the lines to their high state. There are countless RS-485 transceivers that can’t even run 128 nodes, and those signals are DRIVEN both low and high (doesn’t rely on pull-ups). Can anyone elaborate on I2C with large numbers of nodes? Is it really possible to operate 128 nodes on the same I2C bus?
Okay, so now I’ve skimmed the article. Interestingly, they discuss using RS-485 in the future, instead of I2C, but they suggest the part MAX481… If you look at the datasheet for MAX481, you will see that the number of receivers on the bus is limited to 32. There are RS-485 transceivers that can drive up to 256 receivers, but they will need to realize that each receiver typically adds >500uA of stand-by current. With 256 nodes, that’s 128mA; this would be a HUGE load for a battery-powered system.
Seems to me like they could have used the I2C data line for the initial addressing via the p2p serial, and have one less connection to worry about. Or am I missing something?