Identifying human movements from still images or video sequences is really challenging due to different problems like partial occlusion, lighting, blurred appearance, background clutter, etc. Many applications are being introduced these days to learn AI through some apps like Google machine learning platform, no-code AI platform, on-premise AI platform or machine learning playground (ML Playground). Let’s take a look at Robolink’s newest App, the Skeletal Tracking AI App. How can it be used to play the popular Brick Breaker game?

Who is Robolink?

Robolink was established in 2012 in San Diego, California to encourage students to engage with STEM. They offer different courses and camps virtually. In addition, workshops are being introduced if you’re not sure which coding style works for you. You can learn AI with object detection, Teachable Snake or Snake Game, VEX IQ coding, Drone Piloting, etc. But one app is being added to their learning courses like Google’s teachable machine. Let’s take a close look!

What is Skeletal Tracking?

Skeletal Tracking AI analyzes a person’s pose in an image or video by identifying key points on a person’s face and body such as nose, eyes, ears, and joints like shoulders, elbows, wrists, hips, knees and ankles. 

How does it work?

Skeletal tracking works once AI identifies your key points. It will track your movements and determine your pose or gestures. See the sample image below where AI is tracking the nose, eyes, ears and shoulders. 

Skeletal tracking key points
Skeletal Tracking AI App tracking the nose, eyes, ears and shoulders using a webcam.

Applying Skeletal Tracking AI App with Brick Breaker

Brick Breaker is a game where a player must smash a wall of bricks by deflecting a bouncing ball with a paddle on the screen. Normally, arrow keys or touch screen controls are being used to move the paddle side to side of the screen, but Skeletal Tracking AI that you train to recognize your poses and gestures will be used instead in this project to control the paddle.

To get the best outcome of this project, you have to make sure that your poses or gestures can be identified easily from one another when you train your AI when moving the paddle left and right. Otherwise, it wouldn’t precisely follow your movements.

Pre-trained model

 Robolink has a sample model to give you a head start. See the images below:

Sample pose

Capturing Poses and Application

Your computer’s webcam will be used to capture the images of your poses to move the Brick Breaker paddle. Make sure that the place is well-lit so it will easily identify your pose. Step by step process is provided to guide you through. Training the AI would take only a few minutes. Once done, you may now play the game. You can adjust the speed on how fast you want the paddle to follow your movement.

Here’s a sample game using nose to control the paddle:

Skeletal Tracking game

We all know that Google’s teachable machine is now getting popular expanding to its 2.0 version as well with multiple teachable machine projects. Skeletal Tracking AI is also compatible with Teachable Machine. Brick Breaker is not an easy game to play with your movement but it’s fun to give it a try. As you play around with it, you’ll be able to figure out which key point would be the easiest to get a high score.