GO-GO GA-GA

Overview

KiddyKam introduces a revolutionary tool to support parents look after their children. By utilizing an object detection model, KiddyKam can keep track of a baby and follow it around the house. The parent can monitor their baby via the live webcam stream through the VIAM app.

What inspired us

We were interested in merging object detection with robotics and VIAM offered the perfect opportunity to do so. Although none of us are parents just yet (thankfully), busy households may not be able to afford meticulous supervision at all times. During that downtime, KiddyKam will be able to supplement the parent's care.

How we built it

We built the baby detection using VIAM's computer vision and machine learning. We also accessed the point cloud data from the camera to help follow the baby at a safe distance.

Challenges we faced

We primarily struggled with connecting and maintaining a stable connection to the VIAM rover, since the event WiFi got too saturated but our alternative hotspot connections were unstable as well. However, we were able to resolve this by connecting VIAM's router to ethernet and connecting to the router. In addition, when working on the machine learning model, we tried out a single image classification model, not realizing that in order to determine where the baby was, we needed of coordinates. Luckily, we realized this relatively quickly and we made the switch to an object detection model.

What we learned

We were all new to VIAM's platform so we enjoyed learning the intricacies of their software.

What's next for KiddyKam

We believe that KiddyKam has tremendous room for further development and exploration. Subsequent versions could include a sound detection feature that utilizes the Twilio API to notify the parent if the baby is crying in another room, or a separate app that allows the parent to turn on and off the robot, and control it if necessary.