-
[논문 읽기] Dex-Net 목차정리논문 읽기/Dex-Net 2022. 3. 30. 08:55
* Dex-Net 3.0 Supplementary File
I. Overview - 개관
II. ADDITIONAL EXPERIMENTS - 추가 실험
A. Performance Metrics - 성과 지표
1) Average Precision (AP).
2) Success Rate.
1) Success Rate
2) Attempt Rate
B. Performance on Known Objects - 아는 객체에서의 성과
1) Planarity-Centroid (PC3D).
2) Spring Stretch (SS).
3) Wrench Resistance (WR).
4) Robust Wrench Resistance (RWR).
C. Performance on Novel Objects - 모르는 객체에서의 성과
1) Planarity
2) Centroid
3) Planarity-Centroid
4) GQ-CNN (ADV)
5) GQ-CNN (DN3)
D. Classification Performance on Known Objects - 아는 객체에서의 분류 성과
1) Planarity-Centroid (PC3D)
2) Spring Stretch (SS)
3) Wrench Resistance (WR)
4) Robust Wrench Resistance (RWR)
E. Failure Modes - 실패 사례
1) Imperceptible Objects
2) Impossible Objects
III. DETAILS OF QUASI-STATIC SPRING SEAL FORMATION MODEL - 준정적 흡착 모델 세부사항
IV. SUCTION CONTACT MODEL - 흡입 접촉 모델
A. Computing Wrench Resistance with Quadratic Programming
B. Derivation of Suction Ring Contact Model Constraints
1) Friction Limit Surface
2) Elastic Restoring Torques
3) Vacuum Limits
4) Constraint Set
C. Limits of the Soft Finger Suction Contact Model
V. GQ-CNN TRAINING - 네트워크 훈련
VI. ENVIRONMENT MODEL - 환경 모델
A. Details of Distributions
B. Implementation Details
'논문 읽기 > Dex-Net' 카테고리의 다른 글
[논문 읽기] (2017) Learning Deep Policies for Robot Bin Picking by Simulating Robust Grasping Sequences (0) 2022.03.28 [논문 읽기] (2017) Dex-Net 2.0 (0) 2022.03.22 [논문 읽기] (2016) Dex-Net 1.0 (0) 2022.03.22 [논문 읽기] (2019) Dex-Net 4.0 : Learning ambidextrous robot grasping policies (0) 2022.03.18 [논문 읽기] (2018) Dex-Net 3.0 (0) 2022.03.11