Overview

This page showcases experimental videos of ClutterDexGrasp performing dexterous grasping in cluttered scenes with different object counts: 4, 6, 8, 11, 15, 16, 18, and 20 objects. For scenes with 4, 6, and 8 objects, we provide videos from two distinct setups to demonstrate robustness across varied configurations.

4 Objects (Scene 1)

1
2
3
4

4 Objects (Scene 2)

1
2
3
4 (failed)
5

6 Objects (Scene 1)

1
2
3
4 (failed)
5
6

6 Objects (Scene 2)

1
2
3
4 (failed)
5
6 (failed)
7
8

8 Objects (Scene 1)

1 (failed)
2
3
4
5
6
7 (failed)
8
9
10

8 Objects (Scene 2)

1
2
3
4
5
6
7 (failed)
8
9
10

11 Objects

1
2
3
4
5
6
7 (failed)
8
9
10
11 (failed)
12
13

15 Objects

1
2
3
4
5
6
7
8
9 (failed)
10
11
12
13
14
15
16