How to choose a delivery robot for a restaurant
Pudu, Keenon, different models and configurations. What to decide before requesting a quote.
A delivery robot is the “lightest” entry into robotics for a business. 2—4 week deal cycle, $10—30K per unit, plug-and-play in most cases. But even here there are several decisions to make before requesting a quote.
This is what we ask the client in the first conversation to pick the right model.
TL;DR
- Measure aisle width between tables — minimum 90 cm. Without that, no delivery robot will pass.
- Pudu and Keenon are the two main vendors; differences are in design, tray count, interface language.
- Pilot units are worth taking before the main order — you'll see real guest flow and integration quirks.
- Marketing effect is real but not a guarantee of “+20% traffic”; treat it as a bonus, not the basis of payback.
When a robot actually fits
Delivery robots work well in restaurants with even floors, clear layout, aisles from 90 cm and traffic from 150 guests per day. That's the format of a mid-sized restaurant of 60—150 seats.
Poor fit: small cafés with under 80 daily guests (the robot will idle), narrow lofts with aisles under 80 cm, multi-floor venues without a lift, restaurants with very high guest density at peak (the robot loses speed).
For chains and food courts — separate story: the same model can work great in one restaurant and badly in another due to layout. So a pilot in the first venue is a required part of the process.
Pudu or Keenon
Pudu and Keenon are the two largest vendors in the niche. Technologically they're close: lidar + cameras, multi-tier trays, POS integration, voice prompts.
Differences are in the details. Pudu BellaBot — friendlier design with an emotional display, better for marketing. Pudu KettyBot — more compact, fits the host role at the entrance. Keenon T8/T9/T10 — more “technical” design, slightly higher payload, stronger in Asian restaurants thanks to historical presence.
For Russia and CIS, Pudu is usually better localised (Russian interface). For UAE, both work; the choice tends to be by design and POS compatibility.
What else to check
Beyond the model, look at three things. First — POS system. If you run iiko, R-Keeper or Toast, you need a specific integration; some POS versions are supported out of the box, others need development.
Second — charging and infrastructure. The robot needs a charging dock, which means space and a power outlet outside the guest path. If the restaurant has complex wiring — factor in extra electrical work.
Third — staff training and UX. The robot doesn't replace waiters 100%; it takes part of the load. If staff perceive it as a threat, efficiency drops. Build training and positive positioning into the launch plan — this is a critical part of the project that often gets underestimated.
Marketing effect
Many chains buy delivery robots not so much for savings as for marketing: TikTok, Instagram, food-guide mentions. It works — but exactly once and in the first 2—3 months.
After that the robot becomes background and guests stop posting. That's normal — bake it into your expectations. Base payback should add up on operational savings (part of a waiter's shift) even without the marketing bump. If the case rests only on “guests will come because we have a robot” — that's not a plan, it's a hope.
Want a concrete estimate for your case?
The guide answers “how” and “why”. Concrete “how much and how long” is a separate conversation. Start with the 2-minute audit or just write to us.