Tapster’s robots are built to poke touchscreens | Hacker News Reader

Tapster’s robots are built to poke touchscreens

techcrunch.com

64 pointshugs24 comments

Looks like a fun project, of course I wonder how much it would cost for 10 to 15 robots so that you could build your own little mobile ad click fraud farm.

I guess 100$ for Prusa Mendel 3D printer frame with steppers and controller and 100$ for some no-name tablet. Is this clicking thing profitable?

No, that's why most people use the robots instead for quality control / testing of the products they make.

A company near by uses big industrial robots for credit card reader testing. I am just curious what grey applications are possible.

Just out of curiosity, since the creator seems to be on this thread, I have a question:

I'm sure there's an optimal "taps per second" number, where once you go over that threshold, the screens are unable to recognize the distinction. Can you share those numbers if you have one? Super interested.

I'm not the original poster but the touch screen controller on a number of ST micro based kits I've played with 'debounces' the press to 10mS selectable down to 5mS. So figure slightly less than 100 to slightly less than 200 taps per second. That can also be affected by multitouch settings as well.

(Tapster creator here) I've gotten my bots up to 800/taps per minute. I haven't seen / gotten to a higher upper limit, yet, where it is unable to recognize the distinction. However, what I have seen on iOS is after a few hundred taps, the taps stopped being recognized in that spot on the screen. Don't know if that's a bug or a feature of iOS to ignore touches at some point.

In the context of automation (and click farms) if you are not testing the touch screen obviously, I think better approach can be by man in the middle attack with touchscreen and board. Or even emulate screen interface totally. I dont know if has anyone tried that before

The point is, sometimes you do want to test the full environment including the touch screen. For example, you may want to check if the touchscreen locks up after a number of taps, if your OS/input parsing layer can cope with the jitter (a finger, in contrast to MITM-injected events, will almost never hit the same x/y coordinates in a row), or you want to check against real production hardware/software and not against something that's been modified for testing in any way.

Actually, Samsung and Trustonic have a hardware-protected Trusted User Interface (TUI) that protects against man-in-the-middle attacks [1]. I work at Rivetz and we have developed software that displays in the TUI. Tapster looks like a great solution for automated testing for our environment.

[1] https://www.trustonic.com/news/blog/benefits-trusted-user-in...

I always wanted to build a Lego robot to do UI automation on our Windows apps and embedded devices. Point a camera at the screen, recognize screen elements and then let the robot do the clicking.

Do it!

These are rather fun. Our company has a more expensive oneoff solution, a traditional six-axis robot which we use for testing chip-and-pin integration by pressing the buttons on the pinpad.

Neat! I always wondered how hard it would be to build a machine to play those reaction-based ticket and prize redemption games.

Are they actually reaction-based or do they work like carnival / fair games where they're actually completely rigged and only pretend to involve reaction timing as a lure?

Ugggggggghhhgh. The repercussions of this, I don't want to think about.

Tapster creator here. Which repercussions?

The first thing I thought about was click-fraud.

Anything else?

If I may, there is a "niche", but most probably economically valid market in professional digital forensics and/or data recovery.

The first (obvious) use is tapping access codes or swiping patterns to unlock a device (when/if the number of attempts is resettable [1]).

The second is to document contents, on some phones (unlocked but for which there is no available imaging/copying solution or for specific apps for which there is no external reader), the current procedure is (say a communication app) to manually have a message on the screen, take a photo of it, swipe to the next one, take another photo, etc.

See as an example/reference: https://www.forensicfocus.com/Forums/viewtopic/t=15977/

[1] Since the reset in many cases is through a power reset, an accessory "push button tool" will be needed.

Click fraud on desktop is obviously much easier than on mobile.

Even on mobile, ideally (for the fraudster) it would be done in an emulator. There are probably tricks that can be done to detect emulators though. I would consider those flaws in the emulator that should be fixed.

Click farms are often a room full of phones on racks with cheap labor:

https://www.zerohedge.com/news/2017-05-11/look-inside-chines...

I don't think they're primarily doing "click fraud" for ads so much as account creation (for spam), captcha defeats, and social media manipulation.

Maybe that's all click (tap?) fraud though.

Also... I'm not sure how Tapster compares to click farms on pricing. Aces them on legitimacy though. So... I don't think a Fortune 500 company is going to send their in development app over to Russian mobs for final bug testing. So they might be same thing but very very different market segments.

I think the cat and mouse game of finding an emulator aberration vs fixing the emulator aberration is waited toward the finding side. Especially for complex systems like current mobile devices, and when they are often heavily locked down like iOS.

At some point the effort to develop a good enough emulator is more expensive than just buying the physical devices.

But another option beyond emulator vs physical devices is physical devices with physical tapping (Tapster) vs physical devices with emulated tapping. That last option seems to be what is used in your video. It might be too expensive to build an emulator, but it is cheaper to (jailbreak? and) emulate touch inputs than to pay a person or a robot to put in real touch inputs.

Ha, and now it hits me that I used the video without any actual humans... Maybe the cheap humans are falling by the wayside.

Reply on news.ycombinator.com