Modern competitive games have a great deal of anti-cheat software working to make sure you can’t hack the games to get a competitive advantage. [Kamal Carter] decided to work around this by building a physical aimbot for popular FPS Valorant.
The concept is straightforward enough. [Kamal] decided to hardmount an optical mouse to a frame, while moving a mousepad around beneath it with an off-the-shelf Cartesian CNC platform, but modified to be driven by DC motors for quick response. This gave him direct control over the cursor position which is largely undistinguishable from a human being moving the mouse. Clicking the mouse is achieved with a relay. As for detecting enemies and aiming at them, [Kamal] used an object detection system called YOLO. He manually trained the classifier to detect typical Valorant enemies and determine their position on the screen. The motors are then driven to guide the aim point towards the enemy, and the fire command is then given.
The system has some limitations—it’s really only capable of completing the shooting range challenges in Valorant. The vision model isn’t trained on the full range of player characters in Valorant, and it would prove difficult to use such a system in a competitive match. Still, it’s a neat way to demonstrate how games can be roboticized and beaten outside of just the software realm. Video after the break.
I don’t understand why you use a mechanical device to interface with a physical mouse. Why not just replace this with a simple HID enabled MCU like a cheap Arduino with HID support? Then you can send mouse coordinates without any complex moving parts. You could still implement some randomness to emulate a human.
I’m pretty sure you could also mount a small lcd screen under the mouse’s camera sensor and move a grid of pixels around simulating movement. Pretty sure the sensor would pick it up, if you use the right colors.
That’s a pretty cool idea
Only very old mice with red LEDs. Nowadays computer mice CMOS-Sensors are sensitive for IR (800-900nm) only.
then why are they still using red led lighting on it? normal spectral halfwidth of a red LED is nowhere near wide enough for that to be an effective illuminator if what you said is true.
I can’t remember the last time I saw a mouse sensor glow in visible light.
Peter – you’re living in a different world than i am…all the mice in my house, including the one i just bought two months ago, have a visible red led that is so bright that it’s painful if the mouse happens to be lying on its back across the room and pass my eye across its beam. i hope your world will percolate to my world! i’m afraid it’s because i only buy the cheapest mouse though
not trying to argue…just an anecdote. really i’m just complaining about the mouse my kid always carelessly throws in a pile when he’s done with minecraft
Well, the IR ones don’t use red LED lighting, of course, but you have a point – in general, the CMOS sensors themselves are still sensitive to a fairly wide range of light (which peaks somewhere around 600-700nm, but is still reasonably good from 400-850 or so). Many IR cameras use an IR bandpass filter so that they reject unwanted light, but I don’t know if all mouse sensors bother (as long as the mouse is on the table, it should block most other light anyway)
I would think that LCD pixels interact with IR to some degree just as they do visible and UV light, and the mouse is illuminating the LCD’s surface from the front with its own LED. So there may not actually be a problem since s long as your display works when front-illuminated, like the once-common “transflective” LCDs, the venerable classic nokia displays, or the modern Sharp memory displays.
Otherwise, my box of old salvaged optical mice (all with red LEDs) and the somewhat newer mice I use daily (all with IR) tend to generally support your statement, though judging by the other comments it is still possible to get a (cheap?) mouse with a RED led.
I am pretty sure that has already being done using a rPiPico. I wonder if this is going to lead to requirind “signed” peripherals to play competitevely and if people will start spoofing those too.
In a money tournament it already should require fully certified equipment subject to review and disqualification.
This is because the TOC for Valorant specifically forbids this. Using a single real mouse technically doesn’t violate their rules.
Possibly because the output of the mechanical linkage would produce enough noise to look real, whereas a rigged HID MCU with no real components would instantly change the aimpoint to a new coordinate, or else directly interpolate in a perfect line from the previous point to the new one, and that might be detectable by anticheat. Or maybe it just looks cooler for the video and gives him something to do that isn’t typing in a console window.
You can also find or edit a final score screen and post in your name and tweet ‘LOOK I WON’
All without being a giant turd towards actual gamers.
You can even say it was an edit, and get praise from a certain nation leader nonetheless.
Or use a graphics tablet to improve.
https://www.youtube.com/watch?v=IjKqjO9MXVQ
Is it wrong to assume that through technology, competitive gaming can never be fair, and most definitely be abused? If that is the reality then it is what it is for want of a better expression! At the limits of detection, it could be argued that an action was taken faster than humans can react, but that is just not true, because sometime humans make a situational predictions that eliminates processing time. In short, online gaming cannot from this point ever eliminate cheating!
Rebirth of the LAN party just to keep from having to play against millions of cheaters or bots from the other side of the globe. Physical presence isn’t a perfect anticheat but it is fairly effective.