hardware Internet Latest news Technology Reviews

Arduino, Accelerometer, and TensorFlow Make You a Real-World Street Fighter

(*15*)A question: must you’re controlling the antique on-line sport Side road Fighter with gestures(*13*), aren’t you merely, you know, facet highway fighting?

(*15*)That’s a question [Charlie Gerard] is going to want to tackle should her AI gesture-recognition controller experiments take off. [Charlie] put together the game controller to be told further regarding the dark arts of device learning in a a laugh and tasty manner.

(*15*)The controller consists of a battery-powered Arduino MKR1000 with WiFi and an MPU6050 accelerometer. Held inside the hand, the controller streams accelerometer data to an external PC, taking footage the characteristics of the motion. [Charlie] skilled three different moves – a punch, an uppercut, and the feared Hadouken – and captured plenty of examples of each. The raw data was once as soon as massaged, reworked to Tensors, and used to train a manner for the three moves. Preliminary checks seem to art work well. [Charlie] moreover made an web fashion(*13*) that captures motion from your smartphone. The demo is outlined inside the video underneath; sadly, we couldn’t get more than three Hadoukens in previous to crashing it.

(*15*)With most device learning undertaking seeming to be aware of telling cats from dogs, this is a refreshing change. We’re seeing a variety of offbeat device learning tasks at the moment, from cryptocurrency wallet attacks(*13*) to a semi-creepy workout-monitoring health club digital camera(*13*).

(*15*)[embedded content]