Skip to content

Customer use case: a global hotel group ​

The demo data models a hospitality operator: one hotel group running a mixed fleet of service robots in 20 hotels on five continents. Hospitality is a common case for robot fleet orchestration: robots from several vendors share lobbies, corridors and elevators, and operations teams need one view across every property.

How the simulation works explains how this fleet was designed and generated. The same file drives both the simulator and the database seed: ingestion-service/cmd/simulator/hotel-fleet.json.

Hotels ​

RegionHotels
EuropeMunich Central, Berlin Mitte, London Canary Wharf, Paris Opera, Madrid Gran Via, Milan Porta Nuova, Zurich Airport
North AmericaNew York Midtown, Chicago Loop, San Francisco SoMa, Toronto Downtown
Latin AmericaMexico City Reforma, Sao Paulo Paulista
Middle East and AfricaDubai Marina, Cape Town Waterfront
Asia PacificSingapore Marina Bay, Hong Kong Central, Tokyo Shinjuku, Seoul Gangnam, Sydney Circular Quay

Each hotel has between 33 and 80 robots, 1,000 in total. Every hotel stores its IANA time zone, because robot activity follows local time.

Robot types ​

TypeShareExample tasksRhythm
Cleaning36%scrub_lobby, vacuum_corridor_f12, vacuum_ballroomBusiest at night
Delivery32%pickup_kitchen, deliver_room_1204, ride_elevator_bank_aBusy from 07:00 to 23:00
Room service16%restock_minibar_f7, ride_elevator_bank_bBusy from 07:00 to 23:00
Reception16%greet_lobby, guide_to_elevator, checkin_assistAlmost always on during the day

Robot ids read as {hotel}-{type}-{number}, for example muc-dlv-07 is delivery robot 7 in Munich.

Behaviour the simulator reproduces ​

  • Robots stay within about 75m of their hotel and move only while active.
  • A task lasts 20 to 120 seconds. When it ends, the robot picks a new task or goes idle, weighted by the local hour.
  • Delivery and room service robots ride elevators for about 30% of their tasks. Elevator rides are a report of their own, because elevator integration is where hotel fleets most often stall.
  • Batteries drain while active and robots dock to charge below 15%, returning to work at 95%.
  • About one robot in a hundred is faulted at any time; a fault lasts 30 seconds to 4 minutes.

Questions the dashboard answers ​

QuestionWhere
Where are my robots and which hotels have problems right now?Overview map: amber clusters contain faulted robots
How available is each hotel's fleet?Availability report
Which robots need a charger before they stop?Low battery report
What is broken and what was it doing?Open incidents report
How much work did each hotel get done, and how often do robots use elevators?Tasks and elevator rides report
What did one robot do in the last hour?Robot page: battery chart and task list
How do I add, retire or reassign a robot?Robots page and robot page: create, edit, delete