Meet Qiro, the AI Chat Bot That Can Engage, Speak and Play with Your Child

Augustinus Nalwan's bot uses object detection, a game engine, an Amazon Polly, and a Selenium automation framework running on a Jetson Nano.

Qiro recognizes the panda Dexie is holding. (📷: Augustinus Nalwan)

Being stuck in isolation can be challenging, especially if you have rambunctious toddlers at home. It can be challenging to keep them engaged and avoid sitting them in front of a TV all day. One dad approached this problem by making a smart AI that can play with and engage his two-year-old son, Dexie. Qiro is a chatbot built by Augustinus Nalwan that has the ability to speak, recognize a toy, and play with your child all via your TV screen.

Qiro, which is a combination of question and curiosity, is a dog chatbot, inspired by Nalwan’s late dog, Pepsi. Powered by an NVIDIA Jetson Nano, Qiro is equipped with an Amazon Polly to help it speak and a Sony IMX219 camera to help it see. Nalwan trained the system’s recognition feature by repeatedly showing it a human face and three of Dexie’s toys using AWS EC2 Deep Learning AMI. The bot is able to identify not only Dexie, but the toys he’s holding up, speak with Dexie, play games, and even search and play videos from YouTube when it’s appropriate using the Selenium script.

Next, he moved on to Qiro’s appearance. It wasn’t enough to just have it look like a dog, Nalwan wanted it to have idle animations. He achieved this by building a skeletal animation system (SAS) and applying it to the sprites of Qiro’s eyes, ears, tail, and other parts he wanted to move. He also built a fidget animation systems based on keyframe animation. All of this is powered by a game engine called Arcade, which is a Python framework.

The system isn’t perfect – it detects the right toys and finds the right videos 80% of the time – but the initial results are promising. Qiro managed to successfully call for Dexie, identify the airplane he was holding, and pull up an airplane video. Nalwan plans on improving the system with a better camera to eliminate blind spots, install a faster FPS, and add more toys for Qiro to recognize.

You can read Nalwan’s detailed process and check out his GitHub here.

Latest articles
Sponsored articles
Related articles
Get our weekly newsletter when you join Hackster.
Latest articles
Read more
Related articles