Goal-based
Restaurant server robots: matching the runner to your floor plan
All three robots target restaurant tray delivery, but their body widths, tray capacities, and autonomy levels differ in ways that become operational decisions before a contract is signed.
The investigation
5 questions this comparison examines
Which robot fits the narrowest dining-room aisles?
Body width sets the minimum aisle clearance needed for unobstructed navigation. Dimensional differences here are not cosmetic.
Bear Robotics ServiPudu BellaBotKeenon DinerBot T10Servi's 17.5-inch width is the narrowest of the three per manufacturer-published dimensions; BellaBot's 30-inch chassis is flagged in Pudu's own documentation as causing navigation lockups in tight passages.
Pudu's skip-for profile explicitly lists tight table spacing as a disqualifier; BellaBot is documented at 2.5 feet wide, making it the structurally widest unit of these three.
How does tray payload differ across the three platforms?
Payload capacity per trip determines how many units a venue needs to cover peak service volume. Each platform carries trays differently.
Bear Robotics ServiPudu BellaBotKeenon DinerBot T10BellaBot's 8-lb per-shelf payload limit is documented in Pudu trade materials; it is the only unit of the three with a published per-tray weight figure.
Keenon documentation describes DinerBot T10 as supporting parallel multi-tray transport in shift-based protocols, a documented capability neither Servi nor BellaBot explicitly names in the same terms.
What autonomy level does each robot operate at, and what does that mean for staff?
Autonomy classification determines how often staff must intervene and under what conditions. Two of these robots sit at different Ladder rungs.
Bear Robotics ServiPudu BellaBotKeenon DinerBot T10Keenon's documentation explicitly describes a teleoperation fallback when obstacle avoidance fails, which caps classification at Level II per the Robovations Autonomy Ladder; Servi and BellaBot have no such documented fallback and both classify at Level III.
BellaBot's 2-5 stuck episodes per shift in busy venues is documented from Pudu operator experience reports across active deployments; Servi provides no equivalent failure-frequency disclosure.
How much floor-mapping and setup work does each robot require?
Initial setup burden matters for operators with seasonal layouts, pop-up venues, or frequent furniture reconfiguration. Mapping overhead varies across these platforms.
Bear Robotics ServiPudu BellaBotKeenon DinerBot T10Pudu publishes a 4-8 hour venue mapping requirement before BellaBot launch; temporary event setups break navigation and require full re-mapping, per Pudu deployment documentation.
Keenon reports a 1-2 week setup window including layout scanning, handoff-zone marking, and staff training; this is the longest documented onboarding of the three platforms.
How does guest-facing expressiveness compare across these platforms?
Dining venues differ on whether they want the robot to blend in or engage diners. Each platform presents a different guest-interaction posture.
Bear Robotics ServiPudu BellaBotKeenon DinerBot T10BellaBot is the only platform of these three with a manufacturer-designed animated cat-face display specifically marketed as a guest engagement feature; Servi and DinerBot T10 use functional display screens without the animated persona.
Servi's 70 dB and BellaBot's 68 dB noise figures are manufacturer-published specs; Keenon provides no noise-level figure for the DinerBot T10 in available documentation.
In closing
What the evidence shows
Patterns that emerged across the questions above.
Chassis width is the first filter
At 17.5 inches, Servi fits aisles that rule out BellaBot's 30-inch frame; DinerBot T10 sits between them. Floor geometry narrows the candidate list before any other dimension applies.
Autonomy levels differ by one meaningful rung
Servi and BellaBot classify at Level III with staff repositioning on failure; DinerBot T10 sits at Level II because its teleoperation fallback is a documented architectural dependency, not an edge case.
Setup overhead varies by weeks, not hours
BellaBot requires 4-8 hours of LiDAR mapping; DinerBot T10 onboarding runs 1-2 weeks including staff training. Venues with frequent layout changes carry that re-mapping cost on an ongoing basis.
Common questions