Lies We Can Trust: Quantifying Action Uncertainty with Inaccurate Stochastic Dynamics through Conformalized Nonholonomic Lie groups

arXiv:2512.10294 · cs.RO, cs.SY, eess.SY · Submitted 2025-12-11 · Read on arXiv

cs.RO, cs.SY, eess.SY

Submitted: 2025-12-11

Updated: 2026-08-22

Code: https://github.com/UM-ARM-Lab/claps

Project page: https://um-arm-lab.github.io/claps

Terminology

• Calibration — 18× in this paper · explained in 7 episodes
• Conformal Prediction — 15× in this paper · explained in 5 episodes
• Accuracy — 12× in this paper · explained in 1 episode
• Epistemic Uncertainty — 11× in this paper · explained in 9 episodes
• State Space — 11× in this paper
• Empirical Coverage — 4× in this paper · explained in 2 episodes
• System Identification — 4× in this paper · explained in 3 episodes
• Aleatoric Uncertainty — 3× in this paper · explained in 4 episodes
• Conformal Lie-group Action Prediction Sets — 3× in this paper
• Fidelity — 3× in this paper · explained in 2 episodes
• Point Prediction — 3× in this paper
• Stochastic Dynamics — 3× in this paper · explained in 2 episodes
• Uncertainty Quantification — 3× in this paper · explained in 8 episodes
• Interpretability — 2× in this paper · explained in 15 episodes
• Theoretical Guarantees — 2× in this paper · explained in 2 episodes
• Convergence Guarantees — 1× in this paper · explained in 3 episodes
• Decoupling — 1× in this paper · explained in 2 episodes
• Domain Shift — 1× in this paper · explained in 4 episodes
• Monte Carlo — 1× in this paper
• Reinforcement Learning — 1× in this paper · explained in 4 episodes
• Symplectic Euler — 1× in this paper

Sources

• Theoretical Foundations of Conformal Prediction

Related papers

• FMT x: An Efficient and Asymptotically Optimal Extension of the Fast Marching Tree for Dynamic Replanning
• MPCFormer: A physics-informed data-driven approach for explainable socially-aware autonomous driving
• RoboLab: A High-Fidelity Simulation Benchmark for Analysis of Task Generalist Policies
• HRDexDB: A 4D Dexterous Grasping Dataset Across Human and Multiple Robot Embodiments
• APT: Action Expert Pretraining Improves Instruction Generalization of Vision-Language-Action Policies
• Fine-tuning is Not Enough: A Parallel Framework for Collaborative Imitation and Reinforcement Learning in End-to-end Autonomous Driving