Heavy and Bulky Item Handling in Distribution: Two-Person Lift Data

What "bulky" changes

In most warehouse tasks the item is small enough that one worker handles it and the main variable is speed. With a refrigerator, a sofa, a mattress or a flat-pack wardrobe, size and weight dominate. Two people may lift in coordination, a hand truck or dolly may carry the item, and doorways, corners and stacking height decide the route. The question for a humanoid team is whether any of this is feasible for a given platform at all, and what kind of data would even speak to it.

Weight limits are a platform question, not a data one. This page does not state what a humanoid can lift, and it draws no safety or compliance conclusions. The NIOSH manual for the Revised Lifting Equation cited here estimates low-back risk in manual lifting tasks from factors such as horizontal and vertical location, travel distance, asymmetry, frequency and hand-object coupling. Nothing about two-person lifts is drawn from it, and no lifting limit is taken from it. At most, its factor names suggest fields a labeler might record for a bulky move, such as carry distance or whether the lift involved a twist. Compare with sack and bag handling, where the load is heavy but compliant rather than large and rigid.

Task anatomy

1. Assess. The worker looks at the item, its packaging, handles or lift points, and the path to the destination.
2. Plan the move. One or two workers, or a hand truck, dolly, pallet jack or lift-assist, are chosen. Spoken or gestured coordination decides who leads.
3. Lift or tilt. The item is tilted onto a dolly, raised to a surface or lifted together. Grip points vary by item: edges, built-in handles, corners.
4. Move. The item is carried or wheeled along a route, through doors, around corners, up or down a ramp.
5. Place or load. The item is set on a rack, a pallet, into a trailer or onto a truck, sometimes tilted or stacked with protection.
6. Check and protect. Corners are covered, straps are applied, and damage is checked.

Two-person coordination is likely to be the hardest signal to capture, and buyers should not assume existing robotics datasets cover it.

What footage and records can support

Multi-person coordination can in principle be annotated: who leads, handoff points, verbal cues. That annotation can support evaluating a system's understanding of coordination, but it is not a control signal.

Limits: what this data cannot support

None of these assets contains robot joint states, commanded actions or contact forces, and the physical load-bearing that defines bulky handling is exactly what video does not record. Training a humanoid to move large or heavy goods needs robot or teleoperation data, probably with force sensing, and footage, SOPs or recognition data alone do not provide it. See human demonstrations vs teleoperated robot demonstrations, force-torque data and bimanual task demonstrations.

Task-specific gaps:

• Human-to-human coordination. Two workers adapt to each other. A humanoid paired with a person, or with another robot, is a different coordination problem. See the human-to-humanoid embodiment gap.
• Equipment-assisted moves may dominate. Sites may use dollies and lift-assist for exactly this reason; footage may show equipment handling more than lifting.
• Varied item set. Each item class has its own lift points and failure modes, so coverage per class may be thin.
• Hazard and incident sensitivity. Heavy handling produces injury risk, and incident recordings are sensitive.
• Low camera coverage in tight routes. Corners and doorways are where decisions happen and where cameras may be sparse.

Illustrative package: bulky-move clips with coordination labels

# Illustrative only - synthetic values; not an available dataset
package: bulky_handling_v0
source: existing dock and warehouse video, 2 distributors (dist_F1, dist_F2)
clips:
  - clip_id: bk_0033
    item_class: wardrobe_flatpack | appliance_upright | sofa | mattress
    nominal_weight_note: "supplier-reported, arbitrary example"
    crew: two_person | one_person_with_dolly | forklift_assisted
    coordination_events: [lead_signal, tilt_together, handoff_at_door]
    route_features: [doorway, corner, ramp]
    outcome: clean | scuffed | dropped | re-routed
views: [dock_fixed, route_fixed]
excluded: weights borne, forces, robot actions, worker identity

"A few dozen moves per item class" in a pilot spec would be an arbitrary planning input, not a minimum. Requirements depend on architecture, task, embodiment and intended use.

Quality and acceptance criteria

• Crew and equipment labeled. Each clip states who and what moved the item.
• Route visible. Clips should show the full path from lift to placement, not only the lift.
• Item information linked. Dimensions and nominal weight come from master data, flagged as supplier-reported.
• Failure inclusion. Dropped, scuffed or re-routed moves are retained.
• Hold-out. Keep item classes or sites out of training for evaluation; see training vs evaluation datasets.

Who may hold it

Furniture and appliance distributors, home-delivery logistics providers and mattress or flooring distributors often handle bulky goods. Their footage is usually dock, warehouse or vehicle-loading video recorded for security, claims or training. Compare with trailer and container unloading and trailer loading and load building for adjacent tasks, and with ergonomic assessment videos for recordings made to assess lifting.

Permissions and provenance

Claims and incident recordings are the sensitive part of this material: a dropped appliance or an injured mover may be under dispute. Ask whether any footage is tied to a claim or dispute, who can license it, and what notice workers received. Delivery records may carry customer names and addresses; masking them does not settle permissions. Coordination labels are for training and evaluating robots, not for reviewing individual movers' technique. See worker consent and identifiable people. This is not legal advice.

Questions to ask suppliers

1. Which item classes are recorded, and how many moves per class?
2. How are two-person lifts and equipment-assisted moves labeled?
3. Do cameras cover the full route, including doors and corners?
4. Are item weights and dimensions available, and who supplied them?
5. Do any recordings come from incident, claim or ergonomics reviews?
6. Are dropped, scuffed or re-routed moves retained?
7. What notice did workers receive, and who can license the footage?
8. Do records or video show customer addresses or order details?

Discuss your requirements

For bulky goods, describe the item classes, crews and permitted uses to SourceX. SourceX looks for US businesses that may hold suitable material, checks the data and the supplier's licensing permissions, and coordinates the license and delivery. If no existing asset fits, SourceX can scope a commissioned collection of new recordings. A request does not guarantee a match; SourceX follows up about potential sources. Start from the humanoid training data guide or the warehouse hub, and see supply chain and logistics records.

Discuss your humanoid training-data requirements with SourceX

Sources

• Applications Manual for the Revised NIOSH Lifting Equation (DHHS (NIOSH) Publication No. 94-110) — human lifting-risk framework; used only to frame lift attributes, not as robot or two-person-lift data.