Short answer
Trailer and container unloading data records people or equipment emptying floor-loaded cartons and palletized freight at the dock: reading the load, choosing the next carton, pulling it clear and passing it to a conveyor, cart or floor. Distributors, 3PLs and retailers hold dock video and receiving records, and new recording can add views of the carton wall. Video shows no carton weights, stack dynamics or forces, so control data must come from robots.
The deployment problem
Unloading is one of the most physically demanding jobs on a dock, which explains the automation interest. OSHA's ergonomics overview lists forceful exertion, repetitive motion and awkward postures among the risk factors for musculoskeletal disorders (OSHA). The revised NIOSH lifting equation estimates the low-back risk of manual lifting from horizontal, vertical, distance, asymmetry, frequency and coupling factors (NIOSH, 2021). Pulling cartons from a tall wall deep inside a trailer touches every one of them.
The data problem is that the work happens where cameras seldom look. Dock cameras see the door and the conveyor. The carton wall, and the hands working it, are usually out of view, so a buyer typically finds receiving records without the footage that would explain them.
Task anatomy
| Step | What happens | Decision or skill | Typical failure modes |
|---|---|---|---|
| Open and assess | Doors opened; load checked for shifted or leaning freight | Whether it is safe to start and where to begin | Collapsing stack; missed damage |
| Position equipment | Conveyor, cart or floor space set up at the door | Reach and alignment to the cargo face | Conveyor too far; door blocked |
| Choose the next carton | Read the wall of cartons | Which carton is free to move, top-down or by layer | Pulling a load-bearing carton; tumbling stack |
| Grasp and free | Lift, tilt or drag a carton clear of neighbors | Grip point and leverage | Snag; torn flap; drop |
| Transfer | Pass the carton to a conveyor, cart or pallet | Orientation, label side up | Conveyor jam; label hidden |
| Handle exceptions | Crushed, leaking, overweight or unlabeled cartons | Set aside and report | Burst carton; unrecorded damage |
| Clear and verify | Sweep dunnage, confirm the trailer is empty, reconcile counts | Counts against the advance ship notice | Cartons left behind; count mismatch |
Floor-loaded freight and palletized freight are different jobs. Pallets arrive with forklift or pallet-jack work and are covered under depalletizing mixed cases and receiving and inbound checks. Containers and trailers also differ in length, door geometry and how tightly cargo was stowed, so record which one each clip shows.
Which data feeds which part
Reading the wall. The perception problem is separating individual cartons in a tightly packed face with little visible gap. A published analogue is SKU-110K, an annotated dataset of densely packed retail shelves aimed at tightly packed, often identical objects (Goldman et al., 2019). It shows shelves, not carton walls, but it makes the same point: dense, repeated objects need instance-level labels and often depth.
Choosing and pulling. A head-mounted camera on an experienced unloader shows which cartons are taken first and how they are braced, hooked or tilted. A fixed camera looking into the trailer from the door shows the wall change after each pull, which produces a natural labeled sequence of what moved.
Receiving records. Advance ship notices list expected cartons, SKUs and sometimes weights; door scans and damage notes record what arrived. Where cartons or pallets carry a Serial Shipping Container Code, the 18-digit key links the logistic unit to despatch advice data (GS1), giving a join between footage and shipment records. Joined to video, these records give throughput per trailer and exception labels without hand annotation.
Effort and load. Plants that study manual lifting may hold ergonomic assessment recordings with the inputs the lifting equation needs; see ergonomic assessment videos. These describe how severe the task is, not how to do it.
What this data can and cannot support
| Asset | Supports | Cannot support alone |
|---|---|---|
| Dock-door and conveyor cameras | Throughput, flow, exceptions | Seeing inside the trailer; hand detail |
| Head-mounted recording of unloaders | Carton order, grasp points, bracing, exception handling | Carton weight, friction, robot reach |
| Fixed view into the trailer, ideally with depth | Wall segmentation, next-carton choice, stack-state change | Whether a carton is load-bearing from appearance alone |
| Advance ship notices and receiving scans | Carton counts, SKUs, timing, damage labels | Any visual signal |
| Ergonomic and lifting analyses | Task severity, lifting conditions | Robot actions |
Video does not show carton weight, contents, friction between cartons or whether pulling one carton will bring the stack down. Human unloaders judge these by feel and experience. A humanoid needs robot or teleoperation episodes for the pull itself and for recoveries when a stack shifts; see human demonstrations vs teleoperated robot demonstrations. Human pace and posture also do not transfer to a robot's reach and strength; manual load handling and carrying covers the general case.
Capture and labels
For commissioned recording, combine three views: a head-mounted camera on the unloader, a fixed camera at the door looking into the trailer and a camera on the conveyor or cart. Use depth or lidar for the wall view if your robot will perceive it that way. Trailers are dark, dusty and hot in some seasons, so test exposure and lens protection before the pilot. Align every view to the scanner clock. Describe the dock types and freight mix in a request to SourceX before fixing the camera plan.
# Illustrative only — synthetic trailer, carton classes and values
unload_id: UL_site031_0212
equipment: {type: trailer, loading: floor_loaded, door: dock_door_09}
asn: {lines: 42, expected_cartons: 1180, weight_known_per_carton: partial}
cameras:
- {id: head_unloader, view: egocentric, fps: 30}
- {id: door_wall, view: fixed_into_trailer, sensor: rgbd, fps: 15}
- {id: conveyor_end, view: fixed, fps: 30}
carton_classes: [small_light, medium, large_light, large_heavy_flag]
events:
- {t: 0, type: doors_opened, load_check: "leaning stack, right side"}
- {t: 182.4, type: pull, carton_id: cx_0031, class: medium, grasp: top_face_edge}
- {t: 197.9, type: exception, kind: crushed_carton, action: set_aside}
labels: [carton_instance_masks_per_wall_image, next_pull_choice, grasp_point, exception_type]
totals: {scanned_cartons: 1177, exceptions: 4, unloader_id: operator_A31}
excluded: [robot actions, carton weights not recorded, contact forces]
Plan coverage on purpose: carton size and weight mix, stack height, clean, shifted and leaning walls, trailer versus container, packaging type (printed cartons, sacks, wrapped bundles), damaged freight, season and lighting, and several unloaders with different methods. Sack and bag handling is a separate task with its own deformable-load problems.
Supplier types
- Distributors and wholesalers. High inbound volume of mixed cartons, often floor-loaded.
- Third-party logistics providers. Receive for many clients, so the freight on camera belongs to others; see client-owned inventory rights and 3PL providers as data suppliers.
- Retail distribution centers. Store-bound freight arriving floor-loaded.
- Import warehouses and container freight stations. Unload ocean containers; ask whether they record at all.
- Equipment vendors. May hold test footage from conveyor or unloader trials, usually under vendor terms.
The related outbound task, trailer loading and load building, shares suppliers and records but reverses the decisions.
Limits for this task
- Trailer interiors are poorly covered. Existing cameras rarely see the wall or the hands.
- Safety review. Recording near dock plates, forklifts and unstable freight needs the site's approval.
- Conditions degrade images. Dust, glare from the door and dim interiors reduce usable frames.
- Weights are often unknown. Cartons without recorded weights leave effort unlabeled.
- Freight is site-specific. A distributor of tires and one of apparel produce different walls.
- Exceptions are rare. Collapses and bursts need deliberate collection or long archives.
Questions to ask suppliers
- Do any cameras look into trailers or containers, and from where?
- Which freight types, carton sizes and stowage styles would recordings cover, and are weights recorded per carton?
- Are unloads floor-loaded, palletized or both, and in trailers, containers or both?
- Can video be aligned to door scans and advance ship notice lines, and what is the clock offset?
- Are damaged, shifted and collapsed loads kept in the archive?
- Whose freight appears, and have shippers, carriers or clients approved AI training use?
- Do drivers, carrier staff or temporary workers appear, and what notice applied to them?
- Could the site host head-mounted recording on selected inbound shifts with safety sign-off?
Permissions and provenance
Dock footage can show carrier drivers, agency workers, shipper names and purchase-order numbers on cartons, and consignee goods. The operator, the shipper and the people on camera each hold different rights; see worker consent and identifiable people and confidential business information in workplace footage. Redacting labels reduces exposure but does not settle whether a clip may be licensed.
Discuss your requirements
If you are scoping an unloading program, tell SourceX the dock types, freight mix, views and permitted uses you need. SourceX looks for US businesses that hold suitable footage and receiving records, 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 warehouse and fulfillment hub or the humanoid training data guide. The catalog entry for supply chain and logistics operations records and the logistics buyers page give wider context.
Describe the workplace demonstration data your robotics project needs
Sources
- Ergonomics - Overview — ergonomic risk factors for musculoskeletal disorders.
- Applications Manual for the Revised NIOSH Lifting Equation — factors in the manual-lifting risk estimate.
- Precise Detection in Densely Packed Scenes — dense-object detection precedent.
- Serial Shipping Container Code (SSCC) — logistic-unit identifier that links to despatch advice data.