Skip to content

feat: T/F/C/DG/PR-MPPI (33~37번째 변형) 5종 구현#20

Merged
Geonhee-LEE merged 2 commits intomainfrom
feature/advanced-mppi-variants-2
Mar 21, 2026
Merged

feat: T/F/C/DG/PR-MPPI (33~37번째 변형) 5종 구현#20
Geonhee-LEE merged 2 commits intomainfrom
feature/advanced-mppi-variants-2

Conversation

@Geonhee-LEE
Copy link
Copy Markdown
Owner

Summary

  • T-MPPI (33rd, arXiv:2412.17118): Transformer 기반 초기화 학습 → 샘플 효율 50-75% 향상
  • F-MPPI (34th, arXiv:2506.14855): Riccati 피드백 재사용 → 계산 75% 절감 (reuse 모드 ~0.01ms)
  • C-MPPI (35th, arXiv:2412.09777): 중첩 MPPI 비상 계획 → 모든 상태에서 안전 탈출 보증
  • DualGuard-MPPI (36th, arXiv:2502.01924): HJ 안전 가치 함수 → soft/hard/filter 3모드, MinClearance 최고
  • PR-MPPI (37th, arXiv:2601.02948): 입자 기반 파라미터 학습 → wheelbase 60% mismatch에서 RMSE 35% 개선

Test plan

  • 140 new tests (28×5 variants) all passing
  • 1793 total tests passing (no regression)
  • 5 benchmarks with 4 scenarios each
  • python -m pytest tests/ -v --override-ini="addopts="

🤖 Generated with Claude Code

Geonhee-LEE and others added 2 commits March 21, 2026 11:27
- T-MPPI (Transformer): 학습 기반 초기화로 샘플 효율 향상
- F-MPPI (Feedback): Riccati 피드백 재사용으로 계산 75% 절감
- C-MPPI (Contingency): 중첩 MPPI 비상 계획으로 안전 보증
- DualGuard-MPPI: HJ 안전 가치 함수 + soft/hard/filter 모드
- PR-MPPI (Parameter-Robust): 입자 기반 파라미터 학습 + robust 제어
- 140 tests (28×5), 5 benchmarks, 1793 전체 테스트 통과

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@Geonhee-LEE Geonhee-LEE merged commit 7778573 into main Mar 21, 2026
1 of 3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant