Thread by @oulasvirta

Our 5th paper (cond) accepted to #chi2020 is "Optimal Sensor Position for a Computer Mouse"

If you turn your mouse upside-down, you'll see its position sensor. WHY is it where it is? We present a method for optimizing its position in task and improve user performance by 4-20% 1/

One might think that the mouse is studied through-and-through, being as well-established and old as it is.

But it's not!

Previous work has looked at everything from transfer functions to weight and shape, but the position of the location sensor has been overlooked 2/

Earlier and only published work was done with a ball sensor, and concluded that "moving the ball from back to front for better feel and performance", which we show not to be true universally

[Verplank and Oliver] "Microsoft mouse: testing for redesign"
3/

Here's a demonstration why sensor position matters: exactly the same hand motion results in very different cursor motions depending on the sensor position

Why? The hand rotates while moving a mouse.
4/

We first created mice with movable sensors, one that moves the sensor physically on a sliding track, and another that has a VIRTUAL sensor position model based on readings of two sensors.

When tracking error is small, the virtual sensor is approx. as good as the physical
5/

We then developed 1) a calibration method and 2) a human-in-the-loop optimization approach that allows identifying a user's best sensor position within an hour.

You can improve pointing performance by 4-20% depending on what the starting position was.
6/

We believe that high-performance users like gamers can benefit from this method.

And it's cheap: implementing the virtual sensor model requires just 1 extra sensor, calculation in software-side are straightforward.
7/

Thanks to lead PI @sunjun and our colleagues Thomas van Gemert and Byungjoo Lee / KAIST. 👌
8/

@sunjun @sunjun_kim @byungjoolee85 Thomas van Gemert