Seeking Quality Diversity in Evolutionary Co-design of Morphology and Control of Soft Tensegrity Modular Robots

Aug 27, 2026

Designing high-performing soft modular robots is challenging because morphology and controller interact in complex, hard-to-model ways. Evolutionary algorithms paired with physical simulators offer a practical route to address this coupling and to generate not only fit solutions but also diverse design candidates. This study improves quality diversity in co-evolved Tensegrity Soft Modular Robots for goal reaching and narrow-passage squeezing using MAP-Elites, ViE-NEAT, and DM-ME, showing task-dependent fitness and stronger diversity coverage with DM-ME.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages9

Reading Time23 min

AccessFree to download

Abstract

Introduction

Motivation for Soft Modular Robots

Evolutionary Design and Quality Diversity

Focus on Co-optimization for TSMRs

Seeking Quality Diversity in Evolutionary Co-design of Morphology and Control of Soft Tensegrity Modular Robots Enrico Zardini University of Trento Trento, Italy enrico.zardini@unitn.it Davide Zappetti École Polytechnique Fédérale de Lausanne, Switzerland [davide.zappetti@epfl.ch](davide.zappetti@epfl.ch) Davide Zambrano École Polytechnique Fédérale de Lausanne, Switzerland [davide.zambrano@epfl.ch](davide.zambrano@epfl.ch) Giovanni Iacca University of Trento Trento, Italy giovanni.iacca@unitn.it Dario Floreano École Polytechnique Fédérale de Lausanne, Switzerland [dario.floreano@epfl.ch](dario.floreano@epfl.ch) ABSTRACT Designing optimal soft modular robots is difficult, due to non-trivial interactions between morphology and controller. Evolutionary algorithms (EAs), combined with physical simulators, represent a valid tool to overcome this issue. In this work, we investigate algorithmic solutions to improve the Quality Diversity of co-evolved designs ofTensegrity Soft Modular Robots (TSMRs) for two robotic tasks, namely goal reaching and squeezing trough a narrow passage. To this aim, we use three different EAs, i.e., MAP-Elites and two custom algorithms: one based on Viability Evolution (ViE) and NEAT (ViE-NEAT), the other named Double Map MAP-Elites (DM-ME) and devised to seek diversity while co-evolving robot morphologies and neural network (NN)-based controllers. In detail, DM-ME extends MAP-Elites in that it uses two distinct feature maps, referring to morphologies and controllers respectively, and integrates a mechanism to automatically define the NN-related feature descriptor. Considering the fitness, in the goal-reaching task ViE-NEAT outperforms MAP-Elites and results equivalent to DM-ME. Instead, when considering diversity in terms of “illumination” of the feature space, DM-ME outperforms the other two algorithms on both tasks, providing a richer pool of possible robotic designs, whereas ViE-NEAT shows comparable performance to MAP-Elites on goal reaching, although it does not exploit any map. CCS CONCEPTS • Computing methodologies → Evolutionary robotics; Genetic algorithms. KEYWORDS Soft Tensegrity Modular Robots, Co-evolution, NEAT, Viability Evolution, MAP-Elites, Quality Diversity Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions [from permissions@acm.org](from permissions@acm.org). GECCO’21, July 10–14, 2021, Lille, France © 2021 Association for Computing Machinery. ACM ISBN 978-1-4503-8350-9/21/07. . . $15.00 [https://doi.org/10.1145/3449639.3459311](https://doi.org/10.1145/3449639.3459311) ACM Reference Format: Enrico Zardini, Davide Zappetti, Davide Zambrano, Giovanni Iacca, and Dario Floreano. 2021. Seeking Quality Diversity in Evolutionary Co-design of Morphology and Control of Soft Tensegrity Modular Robots. In 2021 Genetic and Evolutionary Computation Conference (GECCO’21), July 10–14, 2021, Lille, France. ACM, New York, NY, USA, 9 pages. [https://doi.org/10.1145/](https://doi.org/10.1145/)[ ](https://doi.org/10.1145/)3449639.3459311 1 INTRODUCTION Soft robots might be one of the key technologies of the future. Indeed, their robustness and adaptive morphology allow them to overcome situations and perform tasks in which traditional hard robots are of limited applicability [18] . For example, they can perform locomotion on rough terrains with risking damage [19], or squeeze through narrow passages [1] . Hence, they can be employed in the exploration of hard-to-access

FAQs

Why is evolutionary co-design important for soft tensegrity modular robots?

Designing soft modular robots is difficult because dynamics are hard to model and morphology-controller interactions are non-trivial. Co-optimization leverages synergy that is missed when optimizing morphology or controller in isolation.

Which evolutionary algorithms are compared in the study?

The work evaluates MAP-Elites and two custom approaches: ViE-NEAT and DM-ME. DM-ME extends MAP-Elites by using separate feature maps for morphologies and controllers and automatically defining the NN-related descriptor.

How do the algorithms compare in performance and diversity for the two robotic tasks?

For goal reaching, ViE-NEAT outperforms MAP-Elites and matches DM-ME in fitness. For diversity measured by feature-space illumination, DM-ME outperforms both other algorithms on both tasks, providing a richer set of designs, while ViE-NEAT is comparable to MAP-Elites for goal reaching.

Aug 27, 2026

Uploaded by Sophia Brooks