Journey to YUNO 04|Why Yuno Had to Feature 360° Pet Tracking
When we began defining Yuno, we kept returning to one question: if we could get only one feature exactly right, which one would most reduce a pet owner's anxiety? The answer was not remote treats. It was not 4K video. It was the ability to keep seeing your pet.
Most pet cameras only see a small area around the food bowl. The moment your cat steps under the table, behind the sofa, or simply out of frame, they vanish. You know they are home, but you do not know where they went, whether they ate, or how their day is unfolding.
That uncertainty became the starting point for Yuno. If Yuno was truly going to bring peace of mind, it could not just watch the bowl. It had to follow your pet.
Fixed cameras wait for pets to come back into frame. Yuno was designed to keep following.The Hardware Reality
Building smooth, silent 360° rotation into a compact feeder was not a small mechanical upgrade. The team had to balance motors, internal wiring, heat dissipation, night-vision LEDs, and industrial design inside a tight footprint.
To cover more real-world home spaces, we also expanded the vertical field of view from 110° to 140°. Yuno does not only see the bowl. It sees the room around it.
Early design studies explored how a rotating camera could feel compact, stable, and approachable.
Dual-axis movement required careful coordination between vertical and horizontal motor systems.The Real Challenge: The Visual Brain
Making the camera spin was only half the battle. A moving camera requires much higher precision from object-tracking algorithms, because the target is no longer sitting inside a fixed frame. The camera has to find, follow, and recover the pet as the scene changes.
Real homes are messy in exactly the ways that matter:
- Cats hide behind furniture.
- Dogs sprint across the frame.
- Lighting changes by the hour.
- Multiple pets move in and out of view.
To make tracking reliable, we trained Yuno on 500+ real pet-owning households and over 100,000 pet activity videos, covering different breeds and body sizes, daylight and low-light scenes, furniture occlusion, partial visibility, multi-pet homes, and high-speed movement.
The result: 95% recognition accuracy in real-world settings.
Yuno's tracking model was tested against the ordinary chaos of real homes: occlusion, multiple pets, low light, and infrared scenes.Tracking That Feels Natural, Not Robotic
Once the algorithm knows where the pet is, the next question is how the camera should move. During testing, we noticed that many auto-tracking cameras suffer from a familiar problem: the lens snaps violently toward the subject, creating a stuttering, jittery view.
We spent months optimizing Yuno's movement so it would not whip around. Instead, it moves more like a human gaze:
- Slow acceleration
- Smooth follow
- Gentle deceleration
If the pet keeps moving, Yuno predicts the motion trajectory and follows along more seamlessly. It is quieter, smoother, and more natural, like a real camera operator rather than a machine.
Movement tuning helped Yuno follow pets without the harsh snap often seen in basic auto-tracking cameras.Multi-Pet? No Problem.
What happens when two pets appear at the same time? How does Yuno know which one to keep following?
We built a continuous visual feature tracking system that analyzes each pet's:
- Body silhouette
- Coat texture and color
- Posture and gait
- Movement trajectory
Even if two pets briefly cross paths or block each other, Yuno works to maintain focus on the same individual rather than randomly switching targets.
Occlusion handling helps Yuno keep identity through short interruptions.
When a pet leaves the frame, Yuno can actively search the surrounding area to re-acquire them.Beyond Seeing: The Shift to Knowing
For us, 360° tracking was never about seeing more for its own sake. It was about knowing more.
What pet owners truly want is not just a video feed. They want answers: Did my pet actually eat today? Are they resting, playing, or acting off? How was their day while I was away?
This led to a bigger challenge. Our vision algorithm needed to move from recognizing a pet to understanding what they are doing. So we added:
- Pose estimation to track body posture.
- Object recognition to identify feeders, bowls, and litter boxes.
- Spatial-temporal mapping to analyze where and when a pet interacts with objects.
Over time, Yuno builds a behavior log that can reveal eating habits, activity frequency, and daily routine changes. Pets cannot tell us, "I do not feel quite right today." But small changes are often already hidden in their behavior.
Not About Gimmicks. About Peace of Mind.
Throughout development, we went through countless rounds of tracking optimization, gimbal tuning, and behavioral testing. We were not doing this to show off. We did it because we wanted to reduce the quiet worry that comes with leaving your pet at home.
In the End, We Realized
True 360° tracking was never really about the camera. It was about making sure that, no matter where your pet goes, you never miss the moments that matter.
Hinterlassen Sie einen Kommentar