Imagine…the board of directors of the world-famous toy constructor manufacturing company “L”.
Although “L” has been named “Toy of the Century” in the past, they are struggling now to boost sales that are no longer follow desirable dynamic.
“What should we do? The world of toys is constantly progressing, kids and their parents want more. Simple toys no longer work.” — Audrey claimed.
“Yes, it’s true! My boys no longer want classical toys. I don’t understand them because in our childhood it was all so different! “ — says Andrew.
Then John stepped into the conversation.
“Recently I’ve found some photos my son has taken. He drowns an action element to our basic brick toys in the photos. Laser beams or archer arrows, based on construction toy set. So, it looked like an animated battle between constructor toy elements and they seemed “alive”. Check this out yourself! — and she showed a picture on her smartphone.”
As everyone looked at the picture, they realized what they have been missing — AR! Augmented reality is a technology that lets people superimpose digital content (images, sounds, text) over a real-world environment. For our company, this may be a new angle of a competition. As we will be able to add life in any constructor setting, we are no longer limited to implement any ideas with our toys in-store promotion, develop and measure toy usage retention, and implement upsell strategies, where customers will need to buy a few more boxes to make a greater story alive.
The following solution can be conceived as a mobile application that can run a set of semantic SLAM algorithms that detect specific objects in real-time processed images and build connections between them. Sparse and Dense Maps approaches can be used based on a specific product use case, such as multiplayer necessity, action visualization, luminosity level, etc.
Of course, such a hypothetical project like this one will need a certain level of customization of existing frameworks and technology, as current semantics SLAM algorithms are capable to work only with popular object recognition, like human faces (Snapchat), arms, cars, etc. So, we would need to collect needed datasets for unique objects, and properly train algorithms to make such things work. Here in UDEV Hub, we are ready to take care of your insane ideas come true!
