Table 1 Comparison of Phy-Q with related physics benchmarks and competitions
From: Phy-Q as a measure for physical reasoning intelligence
Test | Generalization | Categorization | Procedurally | Destructible | Observe outcome | Human |
|---|---|---|---|---|---|---|
environment | to individual | of tasks to | generated | objects | of a desired | player |
physical scenario/s | physical scenarios | tasks/variations | physical action | data | ||
PHYRE12 | ✗ | ✗ | ✓ | ✗ | ✓ | ✗ |
Virtual Tools14 | ✗ | ✓ | ✗ | ✗ | ✓ | ✓ |
OGRE13 | ✗ | ✗ | ✓ | ✗ | ✓ | ✗ |
IntPhys 2019 24 | ✓ | ✓ | ✓ | ✗ | ✗ | ✓ |
CLEVERER25 | ✓ | ✗ | ✓ | ✗ | ✗ | ✗ |
CATER31 | ✓ | ✓ | ✓ | ✗ | ✗ | ✗ |
Physion27 | ✓ | ✓ | ✓ | ✗ | ✗ | ✓ |
COPHY26 | ✓ | ✓ | ✓ | ✗ | ✗ | ✓ |
CausalWorld15 | ✓ | ✓ | ✓ | ✗ | ✓ | ✗ |
RLBench32 | ✗ | ✗ | ✓ | ✗ | ✓ | ✗ |
Computational Pool33 | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ |
Geometry Friends34 | ✗ | ✗ | ✓ | ✗ | ✓ | ✗ |
AIBIRDS35 | ✗ | ✗ | ✓ | ✓ | ✓ | ✓ |
Phy-Q (this study) | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
