Ari: The Learning Console

"A gaming console on wheels" is how Ozobot's CEO pitched it, shorthand for a robot that's actually built to teach kids how to code, not to play. Ari needed an operating system built from scratch, one flexible enough to hold three completely different ways to code it (screen-free with Color Codes, on-screen with the Ozobot Editor and Python, and an all-new Apps layer), plus a personality warm enough that a third grader wouldn't mistake any of it for homework.

UI Exploration and Navigation

Ari’s entire interface lives on a 390×390 pixel touchscreen, small enough that any wasted tap becomes a real problem. Early exploration focused on what a student sees the moment the robot powers on, and how to keep three very different coding paths, screen-free Color Codes, on-screen with the Ozobot Editor and Python, and the new Apps layer, from feeling like three different robots.

Home Screen & Navigation

This is the interface as it actually shipped: the home screen, and how a student moves from it into app categories, individual apps, or back out to Color Codes, without hunting for a menu. Each app works like something on a phone or tablet, a self-contained set of lessons rather than a single screen, Life Cycles, for example, runs four lessons built around picking an animal. The annotated version below adds the details that make it work day to day: the full Color Codes chart, a battery indicator, and the robot’s own name, since a classroom full of Aris needs a fast way to tell them apart.

The Character System

With the architecture set, the job became giving Ari a personality worth watching: over 140 micro-animations, each running half a second to three seconds, spread across three core states. One covers Ari held in a student’s hands (pink), another covers Ari following a Color Codes track (green), and a third, being picked up by ORA, Ozobot’s robotic arm (yellow), got cut for time when the ship date left no room for the lowest-priority state of the three. This personality layer is technically its own app too, just one built permanently into the OS. The diagrams below map the two states that made it in.

Every behavior started as a plain-language list before it became a diagram: what triggers it, how long it plays, and whether it loops or hands off to something else. That list, mapped out in full, is what turned into the flow chart above, then expanded below.

Line-following was the hardest state to get right, since it’s the one mode where Ari is reacting to the real world instead of a tap on the screen. The breakdown below shows how the robot’s reactions escalate as it speeds up, from a steady cruise to full nitro.

The breakdown above is the animatic for the nitro behavior, mapped out before any animation was finalized. The first video is the finished result, the second shows it running live on actual Ari hardware, not just a render. Every animation ties back to a specific Color Codes entry, including the counters in the third video, where Ari counts down from five after crossing a counter code.