AGIdex
agidex/Capabilities/Manipulation/Dexterous Grasping
SUB-CAPABILITY · MANIPULATION

Dexterous Grasping

48%
vs. HUMAN BASELINE = 100
PartialConf · Medium

Scoring rubric

0–20
Early research
20–40
Narrow benchmark competence
40–60
Strong benchmark · reliability gaps
60–80
Human-competitive in common scenarios
80–100
Comparable to typical skilled adult
100+
Reliably exceeds typical human baseline

Score vs. baseline

Trend · Last 12 months
12mo ago   %
6mo ago   %
Today   %
Δ 12mo   +0

What this measures

Human baseline

Pick up and place common household objects without dropping them.

Human frontier

Manipulate deformable, slippery, and tiny objects at human speed and reliability.

Current state

What works

Rigid objects on clear surfaces; Figure 02 handled 90,000+ parts in BMW production pilot.

Key gaps

Deformable objects, cluttered scenes, objects requiring precise force calibration.

Evidence

3 sources
TechnologyQualitySourceScore vs. baselineScore
Atlas (Electric)vendor reportedbostondynamics.com/atlas
48%
Figure 03vendor reportedfigure.ai/news
52%
Optimus Gen 3vendor reportedtesla.com/AI
46%
Source: paperswithcode.com/task/robotic-grasping · Last reviewed May 2026