Greg Davill Sinks His Teeth Into ShArc: A Geometric Technique for Multi-Bend/Shape Sensing

Davill shows us that the gates to cutting-edge research are within the grasp of all of us.

Tom Fleet
a year agoWearables / Sensors

One topic that I think we could possibly look to cover a bit more over on Hackster News are the types of projects that are born out of research labs, PhD internships and other such programs. Indeed, the number of such articles featured doesn't exactly tip the scales when compared to the volume of awesome projects we see from the maker community! That's not to say I'd rather see any less of the latter, moreso, I want to see more of every type of project!

Perhaps there has been reason for this discrepancy though.

Often, these programs are well funded, and aspire to meet many mighty goals — say for example, a new sensing technology — the pursuit of which will often require some pretty specialized equipment in order to research, refine and reliably reproduce their work, ready for release to the public.

That sort of test gear tends to carry a pretty significant price tag, and not to mention, those looking to develop new types of sensors, or process from first principles are dealing with a much less refined product than the likes of you or I would likely be accustomed to working with — there's usually a few stages of production engineering to go before we are going to be able to get to grips with this great new invention.

Yeah, most of the stuff going on between the cover sheets of research papers, and other such publications, tends to be out of the price range of all but the most well provided for of prototyping labs, and likely will remain prohibitively priced — at least until the process or product described within is able to jump off the pages, and into mass production!

But with the years of growth that have seen the technology sector and market not only massively expand, but also massively diversify — there are now research programs into each and every aspect of how we embrace and interact with this world of widgets — as well as programs looking at the potential ways we have yet to propose for interacting with our peripherals.

The research coming out of these labs is, in my opinion, potentially the sort of research that — if you're reading this — I think will likely resonate with the Hackster audience.

Not only for the areas of research covered — ones that have a will likely have a direct impact on us, and the way in which we wave our arms at our gadgets, but also because they are working with technologies and platforms that are available to people such as you or myself, people who are looking to play along with the paper or publication, in pursuit of their own projects.

Sink your teeth into ShArc!

The work of one such research project caught the eye of Greg Davill recently, when a paper written by Fereshteh Shahmiri and Paul H Dietz was published, after being submitted for the 2020 ACM Conference on Human Factors in Computing Systems (CHI 2020).

This paper goes by the title of "ShArc:A Geometric Technique for Multi-Bend/Shape Sensing,' and proposes a novel contour sensor, comprised of a flexible, capacitive PCB sensor, a suitable capacitance to digital converter, and some subsequent signal processing, allowing a two-layer polyamide FPC circuit to cleverly capture the contours of the shape it is stuck to.

This is fairly significant work in my view, as I feel it is able to overcome one of the primary limitations of previous flexible sensors — the use of finely ordered, interdigitated electrode sets, the construction of which limits the full potential of the design.

Measuring metrology

So from the get go here, it's important to differentiate between the types of surface metrology measurement that can be made.

While this method isn't going to displace the various non-contact, optical methods that exist for determining a surface profile, and the incredible depth of data that these tools can generate, it's not really targeting that sort of shape measurement.

Instead, on a much larger scale, this sensor has a world of applications within fields such as dynamic motion capture, where it could potentially replace, or augment data generated by conventional motion capture camera systems, or perhaps be integrated into high-tech clothing as a robust way of measuring things like stance, pose and gait, when performing various activities.

So how does it work?

Going back to first principles, let's look at how a capacitor works.

(We'll borrow the bare minimum from this handy SparkFun tutorial to get us from a to b in the minimum of time!)

In its most basic representation, a capacitor is a device that is able to store an electrical charge.

Constructed from two metal plates, with a layer of something known as "dielectric" material sandwiched in between, the rough layout of your basic capacitor is shown below.

The amount of charge that can be stored between the two metal plates is a function of the overlapping area of the plates, the distance between them, and the relative permeability of the dielectric material used in the construction of the capacitor.

If you want to create a capacitor capable of storing more charge, the general idea is to use plates of a larger area.

Higher voltage capacitors can wind up being fairly sizable devices, as the higher voltage potential between the two plates can necessitate a greater distance between them — in order to stop potential arc / flash-over, in turn, lowering the amount of charge that can be stored, which in turn, leads to larger plates being required in order to make up that difference.

Different dielectrics lead to different charge and discharge current abilities, along with self discharge rates and long term capacity also being factors that ate commonly affected by dielectric type.

So, rather than teaching you all to suck eggs, why are we going over the classroom basics of capacitor construction? Well, the refresher — as lightweight as it was — covers the key points that allow the ShArc sensor to operate as it does. Let's check it out!

ShArc's FPC sensor construction

The sensor electrode for ShArc is a specially laid out FPC board, that forms a set of eight, differential capacitors, the capacitance value for each being able to vary — positively or negatively — in response to deflection of the sensor, along the axis of measurement.

A single capacitor element can be seen below. In its resting state, with the sensor measurement axis laying flat against a surface, as if we're viewing it from above right now, the two parts of each differential capacitor — shown on the reference strip, will measure approximately equal, as they both are covered by the same amount of plate, on the sliding strip.

Now, forgive the following scary collection of calculus — we've borrowed it from the ShArc paper — don't worry, we're not tearing into this; we're going to use it to visually illustrate how ShArc works.

What the above diagram is telling us is that for a given angular displacement (i.e. an arc), the length of the surface along that arc path, or segment, will depend on the distance from the arc centre.

Makes sense, right? For example, if you're winding an inductor, it follows that the number of turns for a given length of wire would be fewer on a larger diameter core, than for a smaller diameter core.

So, if you have two strips of an equal length, tracking the same arc, and fix them at one end, the two strips will end up sliding over each other, the linder displacement relative to each other will be proportional to the radius of the arc they are following.

(The associated paper is well worth a look, if you can track it down through some sort of... scientific hub... - you'll forgive us for not linking directly)

So, how does this all tie together into a sensor? Well, let's take a look at the graphic below. Here, we can see how the electrodes on each strip vary in position to each other, when the strip is curved.

We can see that as we track along the curved sensor set, that the displacement of the two strips, relative to each other, increases along the sensor. This results in a specific capacitance value of each differential sensor pair, which can be characterised to represent the curve it is currently following.

That's the operation in a nutshell, so why are we covering all this here on Hackster? Well, it's all about accessibility! This research isn't relegated to labs where we'll never see sight of it, until commercialized into a product. Far from it. Davill has shown just how easily we here at home can play along with this project, using the same tools and services that we'd normally look at for our own hobby projects!

He's not only managed to recreate the capacitance to digital converter needed for this application, but perhaps more of note, he's even turned his hand to having a go at the flexible sensor electrodes themselves, all fabricated by the one stop shop, whose services seem to keep on growing— our favorite board fab house, OSH Park!

Clever capacitance to digital conversion

With the system operation and component selection described in the ShArc paper, Davill was able to design a suitable board around the Analog Devices AD7746 24-bit, two-channel capacitance to digital converter IC.

This device is able to resolve changes in capacitance down to 4aF — yes, atto-Farads, with an update rate from 10 to 90 Hz — more than suitable for the sort of range of movements that humans are capable of!

Although hugely capable, the AD7746 won't do much without some control circuitry to toggle it pins.

Davill reached into the jar of family favorites, and came out with the tried-and-tested SAM D21 MCU, from Microchip. This USB-enabled Cortex-M0+ is a great little device for when you need to whip up a custom bit of connected hardware.

Running at 48MHz, with a suite of communication peripherals that can be configured to talk to nearly every embedded interface that exists, it's also cheap enough to throw at projects, with a layout that's simple enough to sit on two layers.

With the addition of some Texas Instruments TMUX1511, four-channel SPST analog switches, the SAM D21 can take care of sequencing the connections between the eight capacitor sets, and the inputs of the AD7746 merely by toggling some I/O pins — a nice and easy solution to the problem.

Get your flex on!

What may be even more interesting of note than the controller board — though it's a lovely layout, characteristic for Davill — are the OSH Park flexible substrate sensor electrode FPC boards that Davill has derived from the figures shown in the paper.

These beautiful, bendy boards are more than up to the task of serving as flexible sensor electrodes, with the sensor strip comprised of a sandwich of two parts, stacked on top of each other!

While we are still waiting on results from our man on the ground Davill, we're sure it's going to be a success. The paper is based on solid principles, and the technologies used are demonstrably within our reach, as hobbyists.

Davill has shown us that with a little bit of light reading, and by calling upon the tools and production services we already know and love through familiarity, it's possible to play along with the published research we so often find ourselves tantalized by.

Perhaps it's a testament to the capabilities we now find commonplace, from products such as the ADD7446 Capacitance converter used here, or the freakin' awesome flexible PCB fabrication that is now made available to us at hobbyist-friendly prices from OSH Park.

Or perhaps it's a sign of a diminishing scope of research available for pursuit. We're getting closer and closer to the blue-sky limits of practical research — imposed by the laws of physics — each and every day. So maybe it's only natural that we start looking at what we can do with these tools, rather than looking to constantly invent a new one.

Davill has set up a thread on Twitter where you can keep an eye on the development of the project — that is — when he gets a spare moment from all the other cool stuff he's working on, and he has also made the FPC electrode board files available over on his GitHub, which might serve to remove some of the air of mystique when designing your own flexible circuits!

Tom Fleet
Hi, I'm Tom! I create content for Hackster News, allowing us to showcase your latest and greatest projects for the world to see!
Latest articles
Sponsored articles
Related articles
Latest articles
Read more
Related articles