Entries to the final round of HackadayPrize are now closed. What does this mean? It means that today is the start of another day of development!

Now, I have the following items in my backlog that have a high priority

- SG-90 servo version of the servo bracket that can be snapped into place
- Modifying the circuitry to improve stability and safety for distribution of the board and case
- An attractive application that demonstrates the ability of "swarm robots to communicate.

(I wish I could shadow myself as the mechanical, circuit, and software person!)

The first step was to implement the software part which had been not ready in time for the contest deadline. We further improved on the previous FaceSync demo to make a cute cheerleader robot.

The rules are simple. When the user strikes a pose that says "hooray!", the robot shouts with joy.

Google MediaPipe provides a variety of ready-to-use recognition models for developers who want to easily use the results of deep learning. This time I used hand position and posture estimation as well as face. This application calculates if the hand position is higher than the top edge of the face to determine if the user is doing a "Hooray! The robots cheer when they detect the rising flag.

Think about the number of robots; wouldn't it be more interesting and energizing to have three robots than one robot answering? And what will be the effect when the number of robots increases to 10 or 20? Isn't that exciting?

There is no need to worry about performance. Since the cheering voice is stored in the M5Stack beforehand, the WebSocket payload is very small. Even if the number of robots increases, there will be no performance degradation.

Both the facial recognition web application and the Stack-chan cheering squad have their sources available on GitHub.

The possibility of "swarm robot interaction" like this application is something I will continue to pursue personally. Hopefully the robot army won't turn against us before then!