Embodied AI Dataset

Logistics Picking and Sorting Dataset

Creator: Changsha Langhui Information Technology Co., Ltd.
Keywords: ,Embodied-intelligence datasets,Robot imitation learning,World-model training,Langhui,AItraining data

20,000+ebisods . Package 6D position . Transmutation belt movement .. Alien processing .. Logistics automation

episodes20,000+

Position Precision<5mm

Bit strut precision<3°

Boxes/software/packages/foam boxes/cylindricals/othera6 Type packages

Standard Frame Rate30/60 FPS

Dataset Overview

Definition：Logistics selection and sorting data sets are smart data sets provided by Long Salang Langhui Information Technology Ltd.. The delivery package capture, encoded, sorted and alienized processing in the storage logistics scene, providing packages 6D position, conveyor belt exercise, cardboard size and broken labels, and supporting training in logistics automation and capture planning algorithms.

Dataset Name	Logistics Picking and Sorting Dataset
Total Data Volume	20,000+ episodes
Data Type	Video, depth map, package 6D position, conveyor belt motion parameters, sorting actions
Video Specifications	1920×1080 @ 30/60 FPS
Delivery Format	LeRobot 3.0 / RLDS / HDF5 / JSON Metadata
License Type	Licensed Data or Custom Projects

Core Data Fields

Each piece of data contains complete structured fields, covering all requirements from basic metadata to professional analysis.

Field Category	Typical Fields
Storage scene video	Video, conveyor belt speed, camera height and angle, light and cover, workspace number
Package 6D position	Central point coordinates, width, rotation matrix, attitude classification (painting/side-up/clip), confidence-building
Capture and sorter labels	Grab points, claws, target-striking, action results, cause of failure, beat time, trajectory key points
Package Properties	Weight interval, materials (paper/package/foam box), broken labels, face sheet information (desensitization), alien marking, fragile marking
Metadata and Compliance	Warehouse code, date of collection, scope of authorization, privacy de-identification mark, data version

Data Sample Showcase

The following is a de-identified structured JSON sample illustrating the typical data format.

{
  "episode_id": "LOG-2026-0067823",
  "warehouse": "wh_south_A",
  "station": "sorting_line_07",
  "video": {"rgb": "rgb_0067823.mp4", "depth": "depth_0067823.mkv", "fps": 60},
  "parcels": [
    {
      "parcel_id": "P-77821",
      "bbox_3d": {"center": [0.45, 0.12, 0.38], "size": [0.30, 0.20, 0.15], "rotation": [0,0,0,1]},
      "category": "carton", "weight_kg": 1.2, "irregular": false
    }
  ],
  "actions": [
    {"time": 1.23, "action": "pick", "parcel_id": "P-77821", "gripper": [0.32, 0.11, 0.42]},
    {"time": 2.85, "action": "place", "parcel_id": "P-77821", "destination": "chute_B3"}
  ],
  "conveyor": {"speed_mps": 0.8, "direction": "left_to_right"}
}

AI Application Scenarios

Logistics robot capture planning

Package 6D position data training in high-speed capture strategies and collision avoidance models.

Smart sorting system

Visual and action data are trained in the separation of foreign-made, soft-packed and non-physical-based robots.

Mesh Optimization

Stable code training and load optimization algorithms for size and weight data.

Damage detection

The kit on appearance and label data trains logistics damage and anomaly detection models.

The storage of digital twins.

The real selection data drive the simulation of storage and the assessment of capacity.

Multi-robots.

Multi-position data support multi-robots deployment and synergetic strategy research.

Frequently asked questions

Overwrite Alien?

Common logistics packages such as card boxes, soft bags, bags, foam boxes, cylinders, irregular packages, etc.

What's the camera frame rate?

Standard capture is 30/60 FPS, customised for 120 FPS collections based on high-speed sorting needs.

How's the position?

6D position is marked by a combination of deep camera and panel, with an error in position <5mm， <3°。

How to Get a Data Sample?

Please contact the Changsha Langhui Information Technology Co., Ltd. DataAssetsAPI Platform.