From CAD to POMDP: Probabilistic Planning for Robotic Disassembly of End-of-Life Products

TL;DR

This paper introduces a probabilistic planning framework using POMDPs for robotic disassembly of end-of-life products, effectively handling uncertainties and deviations from CAD models to improve efficiency and adaptability.

Contribution

It formulates disassembly as a POMDP, derives models from CAD data, and employs reinforcement learning and Bayesian filtering for robust, adaptable robotic disassembly planning.

Findings

Outperforms deterministic baselines in disassembly time and variance

Generalizes across different robotic systems

Adapts to deviations like missing or stuck parts

Abstract

To support the circular economy, robotic systems must not only assemble new products but also disassemble end-of-life (EOL) ones for reuse, recycling, or safe disposal. Existing approaches to disassembly sequence planning often assume deterministic and fully observable product models, yet real EOL products frequently deviate from their initial designs due to wear, corrosion, or undocumented repairs. We argue that disassembly should therefore be formulated as a Partially Observable Markov Decision Process (POMDP), which naturally captures uncertainty about the product's internal state. We present a mathematical formulation of disassembly as a POMDP, in which hidden variables represent uncertain structural or physical properties. Building on this formulation, we propose a task and motion planning framework that automatically derives specific POMDP models from CAD data, robot capabilities, and inspection results. To obtain tractable policies, we approximate this formulation with a reinforcement-learning approach that operates on stochastic action outcomes informed by inspection priors, while a Bayesian filter continuously maintains beliefs over latent EOL conditions during execution. Using three products on two robotic systems, we demonstrate that this probabilistic planning framework outperforms deterministic baselines in terms of average disassembly time and variance, generalizes across different robot setups, and successfully adapts to deviations from the CAD model, such as missing or stuck parts.