Team Ai
Datasetpublic

UMIbenchmark/UMI-Benchmark-v1

UMI-Benchmark-v1 UMI-Benchmark-v1 contains 20,000 real-world robot manipulation sessions across 10 tasks: 4 single-arm tasks and 6 dual-arm tasks. Dataset Summary Category Tasks Sessions Single-arm 4 9,988 Dual-arm 6 10,012 Total 10 20,000 Tasks ID Folder Type Task Sessions T1 single_arm_task1 Single-arm Stack Baskets 3,000 T2 single_arm_task2 Single-arm Trash Bag 1,991 T3 single_arm_task3 Single-arm Stamp Ink 2… See the full description on the dataset page: https://huggingface.co/datasets/UMIbenchmark/UMI-Benchmark-v1.

sourceHugging Facecc-by-4.0updated 13d agoView on Hugging Face
3likes18kdownloads
Dataset Card

UMI-Benchmark-v1

UMI-Benchmark-v1 contains 20,000 real-world robot manipulation sessions across 10 tasks: 4 single-arm tasks and 6 dual-arm tasks.

Dataset Summary

CategoryTasksSessions
Single-arm49,988
Dual-arm610,012
Total1020,000

Tasks

IDFolderTypeTaskSessions
T1single_arm_task1Single-armStack Baskets3,000
T2single_arm_task2Single-armTrash Bag1,991
T3single_arm_task3Single-armStamp Ink2,996
T4single_arm_task4Single-armRemote Storage2,001
T5dual_arm_task1Dual-armPour Beans1,600
T6dual_arm_task2Dual-armPack & Carry1,600
T7dual_arm_task3Dual-armTurntable Pick2,012
T8dual_arm_task4Dual-armMahjong Sort1,600
T9dual_arm_task5Dual-armKitchen Rearrange1,600
T10dual_arm_task6Dual-armPants Folding1,600

Scenario Counts

TASK 01 Stack Baskets / single_arm_task1

ScenarioSessions
Blue Baskets985
Red Baskets1,000
Orange Baskets1,015
Total3,000

TASK 02 Trash Bag / single_arm_task2

ScenarioSessions
White Garbage Bag486
Purple Garbage Bag497
Green Garbage Bag500
Blue Garbage Bag508
Total1,991

TASK 03 Stamp Ink / single_arm_task3

ScenarioSessions
Black Ink + Gray Paper482
Black Ink + Yellow Paper492
Black Ink + White Paper499
Red Ink + Gray Paper500
Red Ink + Yellow Paper505
Red Ink + White Paper518
Total2,996

TASK 04 Remote Storage / single_arm_task4

ScenarioSessions
Remote B + Holder B489
Remote A + Holder B500
Remote B + Holder A501
Remote A + Holder A511
Total2,001

TASK 05 Pour Beans / dual_arm_task1

ScenarioSessions
Red Beans Scene B395
Yellow Beans Scene B400
Yellow Beans Scene A400
Red Beans Scene A405
Total1,600

TASK 06 Pack & Carry / dual_arm_task2

ScenarioSessions
Purple Ball Scene B387
Green Ball Scene B397
Green Ball Scene A403
Purple Ball Scene A413
Total1,600

TASK 07 Turntable Pick / dual_arm_task3

ScenarioSessions
Yellow Can Scene B495
Green Can Scene B500
Green Can Scene A503
Yellow Can Scene A514
Total2,012

TASK 08 Mahjong Sort / dual_arm_task4

ScenarioSessions
Mahjong Combination 2 Scene B389
Mahjong Combination 1 Scene B398
Mahjong Combination 1 Scene A402
Mahjong Combination 2 Scene A411
Total1,600

TASK 09 Kitchen Rearrange / dual_arm_task5

ScenarioSessions
Pink Cup-Plate Set Scene B393
Green Cup-Plate Set Scene B400
Pink Cup-Plate Set Scene A400
Green Cup-Plate Set Scene A407
Total1,600

TASK 10 Pants Folding / dual_arm_task6

ScenarioSessions
Green Pants Scene B391
Blue Pants Scene B399
Blue Pants Scene A401
Green Pants Scene A409
Total1,600

File Layout

Data is stored as tar shards under data/.

text
data/
  single_arm_task1/
  single_arm_task2/
  single_arm_task3/
  single_arm_task4/
  dual_arm_task1/
  dual_arm_task2/
  dual_arm_task3/
  dual_arm_task4/
  dual_arm_task5/
  dual_arm_task6/

Most tasks are sharded by scenario. For example:

text
data/dual_arm_task2/T6_Green_Ball_Scene_A/T6_Green_Ball_Scene_A_chunk_000.tar
data/dual_arm_task2/T6_Green_Ball_Scene_A/T6_Green_Ball_Scene_A_chunk_001.tar

dual_arm_task1 is the only exception in this release. It is stored as 16 sequential session shards:

text
data/dual_arm_task1/session_chunk_000.tar
...
data/dual_arm_task1/session_chunk_015.tar

The scenario counts for dual_arm_task1 are listed above.

Metadata

Metadata files are lightweight index and verification files for the tar shards. They are not the robot trajectory data itself.

text
metadata/<task_name>/
  scene_mapping.csv
  chunk_manifest.csv
  all_sessions.csv
  sha256.txt
  sizes.txt
  sizes_bytes.txt
  lists/

The metadata files provide:

FileDescription
scene_mapping.csvScenario names, session counts, and corresponding data subdirectories.
chunk_manifest.csvTar shard paths, session counts, and first/last sessions in each shard.
all_sessions.csvPer-session scenario membership and relative source paths, where available.
sha256.txtSHA-256 checksums for tar shards.
sizes.txtHuman-readable tar shard sizes.
sizes_bytes.txtTar shard sizes in bytes.
lists/Session path lists used to build each tar shard.

For dual_arm_task1, metadata follows the sequential shard layout:

text
metadata/dual_arm_task1/
  sha256.txt
  sizes.txt
  sizes_bytes.txt

Notes

Each tar shard contains complete session folders. Use the checksum files after download to verify that shards were transferred correctly.