CogWAM/CITATION.cff at main · HorizonRobotics/CogWAM · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
cff-version: 1.2.0
title: "CogWAM: Aligning Semantic Cognition with World Action Modeling via Event-Driven Interfaces"
message: "If you use this software, please cite the paper below."
type: software
authors:
  - family-names: Wang
    given-names: Sen
  - family-names: Liu
    given-names: Liu
  - family-names: Wang
    given-names: Xinjiang
  - family-names: Chen
    given-names: Zequn
  - family-names: Jiang
    given-names: Haoyi
  - family-names: Ding
    given-names: Taojun
  - family-names: Xiao
    given-names: Tingyang
  - family-names: Su
    given-names: Zhizhong
  - family-names: Wang
    given-names: Jie
  - family-names: Zhou
    given-names: Sanping
preferred-citation:
  type: article
  title: "CogWAM: Aligning Semantic Cognition with World Action Modeling via Event-Driven Interfaces"
  authors:
    - family-names: Wang
      given-names: Sen
    - family-names: Liu
      given-names: Liu
    - family-names: Wang
      given-names: Xinjiang
    - family-names: Chen
      given-names: Zequn
    - family-names: Jiang
      given-names: Haoyi
    - family-names: Ding
      given-names: Taojun
    - family-names: Xiao
      given-names: Tingyang
    - family-names: Su
      given-names: Zhizhong
    - family-names: Wang
      given-names: Jie
    - family-names: Zhou
      given-names: Sanping
  journal: "arXiv preprint arXiv:2609.37721"
  year: 2026
  url: "https://www.arxiv.org/pdf/2609.37721"
abstract: >-
  CogWAM couples a vision-language backbone to a dual-width mixture-of-transformers
  that jointly denoises a future visual latent and an action chunk, conditioned on
  an event-driven semantic memory that is refreshed by an explicit KEEP/UPDATE
  decision rather than on every step.
keywords:
  - vision-language-action
  - world-model
  - robot-learning
  - semantic-memory
  - robodojo
license: MIT
version: 0.1.0
references:
  - type: software
    title: "StarVLA: A Lego-like Codebase for Vision-Language-Action Model Developing"
    authors:
      - name: starVLA Community
    repository-code: "https://github.com/starVLA/starVLA"
    license: MIT
    notes: CogWAM is derived from the StarVLA codebase.
  - type: software
    title: "RoboDojo: a robot simulation benchmark"
    repository-code: "https://github.com/RoboDojo-Benchmark/RoboDojo"
    version: 0.2.0
    notes: Evaluation benchmark used for all reported numbers.