A Novel and Accurate BiLSTM Configuration Controller for Modular Soft Robots with Module Number Adaptability

TL;DR

This paper presents a novel bidirectional LSTM controller for modular soft robots that adapts to different module numbers, improving control accuracy and flexibility in complex tasks.

Contribution

It introduces the first LSTM-based control strategy inspired by modular robot structure, capable of handling varying module configurations.

Findings

Controller effective across different module numbers

Validated on both simulated and real robots

Improves control accuracy for modular soft robots

Abstract

Modular soft robots have shown higher potential in sophisticated tasks than single-module robots. However, the modular structure incurs the complexity of accurate control and necessitates a control strategy specifically for modular robots. In this paper, we introduce a data collection strategy and a novel and accurate bidirectional LSTM configuration controller for modular soft robots with module number adaptability. Such a controller can control module configurations in robots with different module numbers. Simulation cable-driven robots and real pneumatic robots have been included in experiments to validate the proposed approaches, and we have proven that our controller can be leveraged even with the increase or decrease of module number. This is the first paper that gets inspiration from the physical structure of modular robots and utilizes bidirectional LSTM for module number adaptability. Future work may include a planning method that bridges the task and configuration spaces and the integration of an online controller.