Restaurant transport robots and a complete table delivery

Read the tray and navigation specifications together

Pudu's BellaBot documentation describes LiDAR and visual positioning, RGBD cameras, tray sensors and a replaceable battery. Its specification lists a maximum total load of 40 kg with a separate 10 kg maximum per tray. The same page labels its 13-hour battery figure as unloaded. These conditions should remain with the numbers. [1]

A total load rating does not authorize placing the entire rated mass on one tray. Nor does an unloaded runtime establish a service duration with dishes, stops and battery wear. Request the test conditions for the intended arrangement.

Before a trial, define tray placement and the permitted loads using the manufacturer's instructions. Use the restaurant's approved handling procedures. Hot liquids and fragile dishes introduce consequences that an empty-tray route cannot test safely or completely.

A route to the table still needs a handoff

Use an illustrative route from the kitchen pass to a collection position beside a table. Staff load the correct dishes and select the destination. The robot travels to the position, someone removes the dishes and the carrier becomes available for another job.

Measure the route with ordinary chair positions and the restaurant's approved access conditions. A path drawn between empty tables may disappear once customers are seated. A robot waiting in that gap can affect both service time and pedestrian access.

1. 1
   Load and identify

   Associate the dishes and table with the intended trip.

2. 2
   Travel

   Record elapsed route time, stops and requests for assistance.

3. 3
   Handoff

   Confirm that the right dishes leave the carrier at the right location.

4. 4
   Return or reassign

   Include the time until the next useful trip can begin.

Compare walking saved with work added

Suppose a hypothetical manual trip needs three minutes of staff walking and one minute of dish handling. A robot-supported trip needs the same minute of handling plus 30 seconds of calling, loading coordination and averaged exception work. Direct staff work falls from four minutes to 1.5 minutes under these assumptions.

Across 24 trips, that is a difference of 60 staff-minutes. It is not evidence that one staff position can be removed. The timing of the trips, table service and other responsibilities determine whether the intervals are useful elsewhere.

Also record the recipient's waiting time and the robot's round trip. A reduction in staff walking can coexist with a longer delivery wait. Neither should be hidden inside a single efficiency claim.

Retain spills, mistakes and manual recoveries

An illustrative service record includes 50 requested trips. Forty-four reach the planned point without help, four require staff to resolve an obstruction and two are cancelled. Report 88% unassisted completion and keep the other six outcomes visible. Count incorrect deliveries or damaged items separately when they occur.

The pilot should include the full service period rather than only quiet minutes. Record who cleans trays, replaces or charges batteries and handles a stopped machine. Food-contact and cleaning procedures remain part of the restaurant's own approved operation.

A useful decision compares this complete record with the existing service method. Customer interaction features may influence the experience, but they do not replace the transport and handling evidence.