Latent Space Reinforcement Learning for Inverse Material Estimation in Food Fracture Simulation

2026-06-15

AI Summary

This work addresses the challenge of directly measuring material parameters in heterogeneous food materials, which exhibit complex internal heterogeneity and non-uniform mechanical properties. To this end, we propose the first universal inverse mapping framework that requires no retraining and generalizes to arbitrary target fracture behaviors. Our approach integrates a neural surrogate model, a normalizing flow-based latent space, and a target-conditioning strategy to efficiently estimate corresponding material parameters in either the original or a reduced-dimensional latent space. By combining proximal policy optimization (PPO) with covariance matrix adaptation evolution strategy (CMA-ES), the method achieves a single-inference time of only 10 ms in an orange-peeling task, yielding a recovery rate of 0.642; further refinement via CMA-ES improves this to 0.828, representing a 23% gain over direct optimization in the parameter space.

Abstract

Realistic visual simulation of food manipulation requires accurate material parameters, yet these are difficult to measure directly and vary across the heterogeneous regions of a single food item. We address the inverse problem of estimating material parameters from a target description of fracture behavior in a non-differentiable continuum damage mechanics simulator. Using orange peeling as a test case, we train a neural surrogate on 2,000 forward simulations and compare Covariance Matrix Adaptation Evolution Strategy (CMA-ES, a gradient-free evolutionary optimizer) with Proximal Policy Optimization (PPO, a reinforcement learning algorithm) across the original 9-dimensional parameter space and two learned 4-dimensional latent representations. Since different oranges have different material properties, a practical inverse system must handle arbitrary targets without retraining. We train a goal-conditioned PPO policy that learns a general inverse mapping: given any target description of peeling behavior, the policy produces a material parameter estimate in a single forward pass (8 surrogate evaluations, approximately 10ms). Operating in a normalizing flow latent space with a shared surrogate evaluator, the goal-conditioned policy achieves 0.642 actual recovery when validated through the simulator, outperforming the original parameter space by 23%. A warm-start extension that initializes CMA-ES refinement from the policy's output further improves recovery to 0.828 with 540 evaluations. These findings provide a practical framework for inverse food physics and lay groundwork for vision-driven material identification from video observations of food manipulation.

Problem

• inverse material estimation
• food fracture simulation
• material parameter identification
• heterogeneous food materials
• non-differentiable simulation

Innovation

• Latent Space Reinforcement Learning
• Inverse Material Estimation
• Food Fracture Simulation
• Goal-Conditioned Policy
• Normalizing Flow