Robotics Research

Overview

Robots are easy. Operators are not.

A warehouse robotics company had reliable machines and an unreliable control panel. Operators juggled fleet status, exceptions and manual overrides across screens designed by five different engineers over four years.

I spent days on the floor in a hi-vis vest, mapped the real exception workflows, and redesigned the console around the ten minutes that decide a shift — when something goes wrong.

Challenge

Designed for demos, used in chaos

The old console looked impressive in sales demos and failed at 2 a.m. with gloves on. Critical actions hid three taps deep; status colors meant different things on different tabs; and there was no way to tell which robot needed a human first.

• Exception queues with no priority — operators triaged by scrolling.
• Override flows that punished hesitation instead of guiding recovery.

Approach

Safety first, then speed

The redesign leads with a single prioritized exception queue: what needs a human, in what order, with one-tap acknowledge. Manual overrides became guided, reversible sequences with plain-language confirmations — no memorized codes at 2 a.m.

Paper prototypes were tested with operators between shifts; the final UI was validated in a simulated outage before a line of production code changed.