Robust 2D Assembly Sequencing via Geometric Planning with Learned Scores

TL;DR

This paper introduces a method combining geometric planning with a neural network to efficiently generate robust 2D assembly sequences, significantly reducing computation time compared to traditional physics simulations.

Contribution

The authors develop a neural network-based approach to predict robustness scores for assembly motions, enabling faster and more reliable planning of 2D assembly sequences.

Findings

Neural network predicts robustness scores with high accuracy.

The approach is an order of magnitude faster than physics simulation.

Successfully applied to two-handed planar assemblies.

Abstract

To compute robust 2D assembly plans, we present an approach that combines geometric planning with a deep neural network. We train the network using the Box2D physics simulator with added stochastic noise to yield robustness scores--the success probabilities of planned assembly motions. As running a simulation for every assembly motion is impractical, we train a convolutional neural network to map assembly operations, given as an image pair of the subassemblies before and after they are mated, to a robustness score. The neural network prediction is used within a planner to quickly prune out motions that are not robust. We demonstrate this approach on two-handed planar assemblies, where the motions are one-step translations. Results suggest that the neural network can learn robustness to plan robust sequences an order of magnitude faster than physics simulation.