Internet of Plastic Things, No Batteries or Electronics Required

Ubicomp researchers seek to blur the lines between hardware and software. Computations (as we typically view them) are modeled as mathematical functions. Typically, computer scientists implement these mathematical functions in software with code. Ubicomp people will certainly do that, but they're also trying to design hardware which performs computation as well.

Significant amounts of work go into making ubicomp systems as energy efficient as possible. Inevitably, you design specialized hardware which doesn't operate under normal paradigms and does indeed appear to be magic.

Take SATURN [1], for example. It's a microphone powered by ambient speech. Energy is harvested though triboelectric nanogenerators (TNGs).

Or take Serpentine [2]. It's a special cord which contains TNGs. Interactions with the cord (pluck, twist, flick, etc.) can be uniquely identified in hardware simply by the generated waveform. There's also no power source here.

The reason things like SATURN and Serpentine come off as magical is because energy-harvesting technologies are not the present standard. We expect to need a power source for our electronics. However, every real-world interaction imparts forces. With clever and efficient hardware, the energy imparted by these forces can be captured and utilized.

That said, this new paradigm also introduces new challenges. How do you ensure one of these devices is off? You can't just unplug the power cable anymore. A possible solution is clearly delineating between the sensor and communication portion and placing a physical switch between these two elements. Since someone here mentioned privacy concerns, I'll just say that any privacy concern will need to be addressed at the hardware/physical level with ubicomp devices.

[1] https://www.sigmobile.org/grav/research-highlight/saturn

[2] https://dl.acm.org/doi/abs/10.1145/3290605.3300775

1. Mechanical actuation generates transmission power. Ok, how is this better than RF induction given that you need to build an antenna into your object anyway?

2. Mechanical actuation changes transmission by tiny (fragile?) gears. This appears to be measuring the wrong thing in their demonstrative example. You don't care about how many times a detergent bottle cap has been twisted. You care about how much detergent is left in the bottle.

All of their examples involve either no sensing or secondary sensing that would be best decoupled from mechanical actuation. A slider could be a super simple variable capacitor. A button requires no signal modification. Leak sensors are typically constructed by simply opening/closing a circuit by direct contact with the liquid. Fluid level sensing is only slightly more involved than leak sensing, and is best done directly, not based on estimated usage.

> Gollakota suggested that e-commerce websites would like to know how a user might be interacting with the objects they sell

Oh, great!

> (after the user has given consent of course)

Of course.

I have always thought of IoT really ultimately being cheap electronics scattered in fields (for ag and gardening) and in buildings, somewhat self-organizing, requiring no sysadmining and ultimately becoming an e-waste. This is more promising, though plastic waste is no better than e-waste.