How we review robots
Our 10-criteria, 100-point scoring rubric — the same one we apply to every consumer, DIY, and industrial robot we test, every time. No exceptions, no paid boosts.
Editorial principles
We buy every robot
No manufacturer loans. No "review units" with strings attached. Every product is purchased with our own funds and never returned for compensation.
We test in real conditions
Floor care bots get cat hair, Cheerios, and threshold strips. Cobots get assembly-line timing. AI toys get a 3-year-old. Real life, not lab benches.
Scores are reproducible
Two reviewers score independently. The published number is the average. If a result varies by more than 5 points, we re-test until we can agree.
Affiliate links never buy scores
Commissions fund the site. We disclose every affiliate relationship. They have zero influence on the score, ranking, or "Best For" recommendation.
The 10-criteria rubric (100 points total)
Each criterion has a weight that reflects how much it should influence the final score in its category. The total is always 100.
| Criterion | Weight | What we measure |
|---|---|---|
| Hardware Quality | 12 | Build materials, durability, component quality, and longevity under real-world use. |
| Software & AI | 14 | Onboard intelligence, update cadence, ecosystem, and adaptability to novel environments. |
| Mobility & Locomotion | 10 | Speed, obstacle handling, terrain, and battery efficiency in real conditions. |
| Manipulation | 10 | Precision, payload, repeatability, and gripper / arm performance. |
| Sensors & Perception | 10 | Camera, LiDAR, IMU, depth, and the perception stack behind them. |
| Connectivity | 6 | Wi-Fi, Bluetooth, 5G, ROS2, and ecosystem integrations. |
| Battery & Power | 8 | Real-world run time vs. claimed, charge speed, hot-swap, and degradation. |
| Support & Warranty | 8 | Documentation, customer service, warranty terms, parts availability. |
| Value for Money | 12 | TCO, repairability, residual value, and 2-year cost of ownership. |
| Safety & Compliance | 10 | Emergency stop, collision avoidance, regulatory certification, and data privacy. |
| Total | 100 | — |
Test environment
We test in three environments: a 1,400-sq-ft lab in Long Island City with controlled floor types and lighting; a 2,300-sq-ft home with two cats, a toddler, and mixed surfaces; and partner small-business sites (a hotel, a 6,000-sq-ft warehouse, a maker space).
Each robot spends at least 30 hours of active use plus 72 hours of standby before we score. Software updates are tracked; any update that changes a score is published as a "Score revision v2".
How scores translate to verdicts
- 90–100 — Best in class. Outstanding, no meaningful weakness. Top pick.
- 80–89 — Excellent. Strong choice with one or two trade-offs. Recommended.
- 70–79 — Good. Solid but not class-leading. Consider if priced right.
- 60–69 — Mixed. Real flaws. Only buy for very specific use cases.
- Below 60 — Not recommended. We won't link to it.
Suggest a robot for us to test
If a robot you'd like reviewed isn't on our list, we want to hear about it. We add 8–10 new products per month.
Submit a robot