Assembly Sequence Predictor

(by Kazuki_Hayashi)

This component predicts an optimal assembly sequence of 3D trusses.
The prediction is implemented by a novel machine learning model trained through reinforcement learning.

Downloads:

427

This component packages a reinforcement learning agent trained for optimizing the assembly sequence of trusses. The assembly order of members predicted by the agent is expected to minimize the temporary supports during the assembly process.

Truss models are converted into a data structure called a graph consisting of nodes and edges, and a novel machine learning method combining graph embedding and reinforcement learning is applied to develop an agent that can efficiently generate a stable assembly path for arbitrary 3D trusses.

## 2022.Apr.27 ##
<Assembly Sequence Predictor component (Assembly Sequence folder)>
･Published the first version.

## 2022.May.1 ##
<Assembly Sequence Predictor component (Assembly Sequence folder)>
･Modified the trainable parameters by re-training.

## 2022.Jun.8 ##
<Assembly Sequence Predictor component (Assembly Sequence folder)>
･Optimized the implementation.
･Display the prediction progress under the component.
<Examples (example GH implementations)>
･Published the first version.