Robot dogs
Cross mixed floors and thresholds, follow a marked route and negotiate a suitable step. Keep remote-controlled and autonomous runs separate. Test only within the maker’s stated limits.
How we test
Real tasks. Every attempt on record.
We are preparing filmed robot tests in everyday spaces around Glasgow: a warehouse, offices, a house, a garden and a converted garage. No ROBODROMO physical test results are published yet.
Our current robot pages are research profiles. Here is the plan for our first hands-on reviews.
Start with one clear job.
For each robot, choose three relevant tasks and define success before filming. Begin with ten attempts per repeatable task. Record one-off setup and build time separately; use a stated protocol for longer battery tests.
Learning kits
Record build time and the time to a first working programme. Repeat a short course or sensor activity and count the help a learner needs.
Desk companions
Check setup, responses to voice and touch, charging and which functions work offline. Keep entertainment separate from claims about care or wellbeing.
Home camera robots
Reach a named room, connect a video call and return to charge where that model supports it. Note blocked routes, lost connections and every rescue.
Workplace robots
When a suitable unit is available, try a defined delivery route or object-handling task. Include loading, unloading, obstacles and the people needed to make it work.
Humanoids
Add hands-on tasks when a suitable unit and safe operating arrangements are available. A demonstration of one configuration does not describe every edition.
Show the work, including the failures.
Film each attempt from start to finish. Keep failed attempts, resets and interruptions in the record. Explain any cuts or missing footage; do not count an edited highlight as a complete trial.
The outcome
Record whether the task finished, time taken, human interventions, control mode, drops or bumps and any safety stop. Record battery use and noise where we have a suitable, stated measurement method.
The human help
Say who sets up the task, drives the robot, gives instructions or rescues it. Report assisted finishes separately from finishes within the task’s agreed rules.
The result
Show completion counts, time and help for each task. Ten attempts are an initial sample, not proof of long-term reliability. Any score must name its task, method and exact configuration.
The conditions
Record the floor, route, objects, lighting and relevant network conditions. Keep earlier results when software or hardware changes, and identify what was retested.
Know exactly what we tested.
Every hands-on review will identify the model, edition, hardware and software; test dates and method version; and whether the unit was bought, loaned or supplied before release.
We will disclose the loan period and return status, any payment or commercial relationship, the access we had to controls and logs, and any help from the maker. Makers may flag factual errors, but cannot approve our verdict.
How we handle research today.
Maker claims, independent reports and our own observations remain distinct. Check the original source, exact edition and date. Missing evidence means unknown, not incapable.
We keep conflicting claims visible and explain meaningful corrections. Availability, service and ownership costs are separate from task performance. Owner feedback sits alongside our assessment and is not silently mixed into a score.
There are no ROBODROMO numerical ratings yet. Filmed results will come first.
Explore the research profiles · Offer a robot for review