Service robots in a fine-dining venue in Dubai
Hybot runs service robots in a fine-dining venue in Dubai. The robots carry orders from the pass to the table and return used dishes to the wash, while Core RMS assigns every task automatically by priority, distance and battery. Performance figures for this deployment are not yet published.
- Venue
- a fine-dining venue in Dubai
- Location
- Dubai, United Arab Emirates
- Robots deployed
- Confirmed with the venue
The challenge
In a fine-dining room the measure of service is how much of the shift staff spend with guests rather than walking. Carrying plates from the pass, taking the bill out and clearing used dishes back to the wash are repetitive, high-volume trips that pull trained floor staff away from the tables they are meant to be looking after, and they get worse exactly when the room is busiest.
The solution
Core RMS takes the carrying. Each floor is mapped from the robot's own points of interest and aligned to a plan image once, so every later map fetch lands correctly. Delivery, collection and bill tasks are then assigned automatically every two seconds, weighted together by priority, distance and battery, and a robot below its battery threshold leaves the pool, charges and rejoins on its own. Staff confirm the human steps; nobody dispatches a robot by hand.
The room, not the robot
Fine dining is judged on attention. The work that erodes it is not the cooking or the greeting — it is the walking: plates out from the pass, the bill out to a table, used covers back to the wash, over and over across a full service. A service robot is worth having in that room only if it absorbs those trips without adding a new job for someone to do.
What Core RMS was set up to do
The deployment follows the same shape as every Core RMS install, because that shape is the product rather than a bespoke build:
- The floor is mapped once. The robot's own points of interest are fetched into a floor plan image and aligned with a persistent affine mapping, so the alignment never has to be redone after a later fetch.
- Tasks are assigned, not dispatched.
DELIVER_ORDER,DELIVER_BILL,GET_ORDERandCOLLECT_DISHESall enter the same queue, and the assignment pass runs every two seconds weighing priority, distance and battery together. - Charging is part of the schedule. Each robot has a battery threshold. Below it the robot leaves the assignment pool, charges and rejoins automatically.
- Every robot reports live. Position, state and battery stream over WebSocket into the dashboard, and per-robot task and distance totals accumulate in the fleet record.
Read how the same loop works across a whole venue on the restaurant solutions page, or see what the hardware integration covers on the robot catalogue.
What we have not published here
This page carries no performance statistics and no photographs. That is deliberate, and it is worth being plain about why.
Core RMS does record the raw material for real figures — tasks completed per robot, distance travelled, time in project, survey responses and average rating. Turning those into a published claim requires exporting them from the fleet record for an agreed window and having the venue agree that the number is fair. Until both of those happen, a figure on this page would be an assertion dressed as evidence, and on a business-to-business site that is worse than silence.
The same applies to the name and to photography. Both are the venue's to give.
If you are evaluating service robots and want specifics rather than a case-study page, request a quote and ask for them directly.
Common questions
- Why is the venue not named on this page?
- We publish a client name only once the client has approved it in writing. Permission has not been granted yet, so the venue is described rather than named. The page will be updated if and when that changes.
- Why are there no performance figures on this page?
- Because none have been measured and verified yet. Core RMS records task counts, distance travelled and uptime per robot, so real figures are available in principle, but we do not publish a number until it has been exported from the fleet record and agreed with the venue.
- Can we speak to a reference customer before buying?
- Ask during the quote conversation. Reference calls depend on the customer agreeing to take them, so we arrange them case by case rather than promising them up front on a marketing page.