Hotel Housekeeping Room Turnover Demonstrations

The deployment problem

Room turnover is a clear humanoid target: the rooms are built for human bodies and the work repeats room after room. It is also one of the longest tasks a service robot could attempt. One departure clean strings together many subtasks across two spaces, each with its own objects, tools, hygiene rules and finished-state standard.

That length shapes the data question. A deployment team needs the task structure and inspection criteria, which hotels document well, and it needs control data for every physical subtask, which no hotel holds. It also has to work around guest privacy, which rules out most recording in occupied rooms.

Task anatomy

Housekeeping standards commonly include ordering rules: top to bottom, dry before wet, clean areas before dirty ones, and separate cloths for toilets, sinks and glassware. Some subtasks can interleave, such as starting bathroom chemicals before making the bed so they have time to work. That partial order is worth capturing explicitly; task graphs from procedures covers how.

Which data feeds which layer

Task structure and inspection. Room standards, cleaning SOPs and supervisor inspection checklists define steps, order and success. They are usually the easiest asset to license and the core of a planner's knowledge for this task; see housekeeping and cleaning SOPs. Occupational task lists help check scope: O*NET publishes task statements and maps them to detailed work activities (O*NET Resource Center).

Long-horizon sequencing. Human video of complete turnovers shows how experienced attendants order and interleave subtasks, recover from interruptions and move between rooms. Published planning work already closes the loop with language feedback: Inner Monologue replanned long-horizon mobile manipulation in a real kitchen using success detection and scene descriptions (Huang et al., 2022). A hotel's inspection items are natural success questions for such a loop. Sequencing data in general is covered on long-horizon task demonstrations.

Physical subtasks. Bed making is two-handed deformable work, and wiping depends on pressure and coverage. Both need robot or teleoperation episodes for control; human video supplies the grasp points, sequences and finished states. See garment folding and textile handling for linen and wiping and surface cleaning for surfaces.

Moving through the room. Turnover is mobile manipulation in tight spaces around beds, carts and bathroom doors; mobile manipulation demonstrations covers what that adds to the specification.

Simulation. BEHAVIOR-1K defines 1,000 everyday activities in 50 scenes, chosen from about 2,000 candidates drawn from time-use surveys and WikiHow, in a simulator that handles deformable objects and liquids (Li et al., 2024). It is household-style, not hotel-specific, but it shows how activity definitions and success conditions can be written down before data is collected.

Capture plan

Record in vacant rooms after checkout or in the training rooms some hotels keep for onboarding. Training rooms give control over setup and consent but lose real clutter; vacant departure rooms are realistic but must be checked for left items and documents before cameras roll. The trade-off is set out in staged vs live operation recording.

Use a head-mounted camera for hands and objects and one fixed wide view per space for whole-body movement and navigation. EPIC-KITCHENS-100 is a precedent for capturing many environments with head-mounted cameras, recording across 45 kitchens and using participants' own narration as the first annotation stage (Damen et al., 2020). For housekeeping, narration helps explain choices the camera cannot, such as why a towel was replaced early.

# Illustrative only — synthetic episode specification for one room turnover
episode_id: turn_site022_room_0417
room: {type: double_queen, layout_id: L3, clean_type: departure}
attendant: attendant_B07 (experience: 2 yr)
cameras: [head_rgb_1080p30, fixed_wide_bedroom, fixed_wide_bathroom]
segments: [assess, strip_beds, trash, bath_chemicals_apply, make_bed_1, make_bed_2,
           bath_clean, dust_wipe, restock, vacuum, final_check]
labels:
  per_segment: start/end time, objects touched, cloth color used, interruptions
  standard_refs: {bed: STD-BED-04.jpg, bath_amenities: STD-BATH-02.jpg}
  inspection: {items_passed: 27, items_failed: 1, failed: ["mirror streaks"]}
  maintenance_flags: ["shower drain slow"]
excluded: robot actions, wiping pressure, lifting force, guest belongings (room checked before capture)

Supplier types

• Hotel groups and management companies. Room standards with photos, training rooms, onboarding video, inspection records.
• Housekeeping contractors. Staff and procedures across several properties, which helps layout diversity.
• Hospitality schools and training programs. Mock rooms and recorded practicals, with learner consent questions.
• Short-term rental turnover services. Varied layouts and less standardized rooms.

Brand standards may belong to a franchisor rather than the property, so ask who can license them; see franchise networks as demonstration suppliers. Standards and SOPs in general are described on the catalog page for licensing SOPs and playbooks. New footage follows the site's separate physical-world path for new recordings of hands-on work, where each opportunity has its own requirements.

Limits for this task

• No forces. Tucking a fitted sheet means lifting a mattress corner; scrubbing needs pressure. Neither shows in video.
• Hidden chemistry. The SOP gives a chemical's dwell time; video shows only a spray. Keep the SOP linked to the episode.
• Vacant rooms miss stayover clutter. Guest belongings, which a robot must avoid disturbing, rarely appear in privacy-safe recording.
• Brand variation. Bed construction, amenity sets and inspection items differ between brands; record which standard applies.
• Two-person work. Some properties make large beds in pairs; decide whether you want team episodes.
• Long episodes. Full turnovers are expensive to segment; budget annotation per subtask.

Questions to ask suppliers

1. Which room types, layouts and brand standards would recordings cover?
2. Are room standards documented with reference photos for beds, bathrooms and amenity placement?
3. Can recording happen in training rooms, vacant departure rooms, or both?
4. How is the room checked for guest items and documents before capture?
5. Are supervisor inspection results recorded, and can they be linked to episodes?
6. Do attendants use color-coded cloths and chemical dwell times, and are those written down?
7. Who owns the brand standards and training video: the property, the management company or the franchisor?
8. Do attendants' releases cover AI training and third-party licensing?

Discuss your requirements

If you are scoping a housekeeping deployment, tell SourceX which room types, subtasks and standards your robot needs to learn. SourceX looks for US businesses that hold suitable material, checks the data and the supplier's licensing permissions, and coordinates the license and delivery. It reviews each requirement and follows up about potential sources; a request does not guarantee a match. Related task pages sit on the workplace task demonstration hub, and the humanoid training data guide covers the other data types.

Describe the workplace demonstration data your robotics project needs

Sources

• BEHAVIOR-1K: A Human-Centered, Embodied AI Benchmark with 1,000 Everyday Activities and Realistic Simulation — everyday activity definitions and simulation with deformables.
• Rescaling Egocentric Vision — head-mounted capture across many environments with narration.
• O*NET Database: Task Statements and Detailed Work Activities — occupational task statements for scoping coverage.
• Inner Monologue: Embodied Reasoning through Planning with Language Models — long-horizon replanning from success detection.