Geometric Red-Teaming for Robotic Manipulation

Published

October 14, 2025

Author(s)

Divyam Goel, Yufei Wang, Tiancheng Wu, Guixiu Qiao, Pavlo Piliptchak, David Held, Zackory Erickson

Abstract

Standard evaluation protocols in robotic manipulation typically assess policy performance over curated, in-distribution test sets, offering limited insight into how systems fail under plausible variation. We introduce a red-teaming framework that probes robustness through object-centric geometric perturbations, automatically generating CrashShapes -- structurally valid, user-constrained mesh deformations that trigger catastrophic failures in pre-trained manipulation policies. The method integrates a Jacobian field–based deformation model with a gradient-free, simulator-in-the-loop optimization strategy. Across insertion, articulation, and grasping tasks, our approach consistently discovers deformations that collapse policy performance, revealing brittle failure modes missed by static benchmarks. By combining task-level policy rollouts with constraint-aware shape exploration, we aim to build a general purpose framework for structured, object-centric robustness evaluation in robotic manipulation. We additionally show that fine-tuning on individual CrashShapes, a process we refer to as blue-teaming, improves task success by up to 60 percentage points on those shapes, while preserving performance on the original object, demonstrating the utility of red-teamed geometries for targeted policy refinement. Finally, we validate both red-teaming and blue-teaming results with a real robotic arm, confirming that the discovered red-teaming failure cases, and corresponding blue-teaming refinement, transfer from simulation to the real world.

Proceedings Title

2025 Conference on Robot Learning

Conference Dates

September 27-30, 2025

Conference Location

Seoul, KR

Pub Type

Conferences

Download Paper

Local Download

Keywords

Red-Teaming, Manipulation, Geometry Perturbation

Citation

Goel, D. , Wang, Y. , Wu, T. , Qiao, G. , Piliptchak, P. , Held, D. and Erickson, Z. (2025), Geometric Red-Teaming for Robotic Manipulation, 2025 Conference on Robot Learning, Seoul, KR, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=960041 (Accessed May 21, 2026)

Additional citation formats