ReMoSo: representations and learning algorithms for highly reconfigurable modular soft robots

Voxel-based soft robots (VSRs) are a kind of robots that are intrinsically modular and reconfigurable, being an aggregation of many simple soft blocks. These features make VSRs potentially suitable for solving a wide range of tasks, provided that their body and brain is tailored to the specific task. Optimization and learning can be the mean through which tailoring is obtained, but current algorithms and representations do not explicitly exploit the intrinsic modularity and reconfigurability of VSRs: a good VSR for a task might be obtained by automatically reusing parts of other VSRs that are good for other tasks. This project will pursue this goal: designing algorithms and representations for the optimization of modular and reconfigurable VSRs. Moreover, for facilitating results and boosting research in this field, software tools and benchmarks will be developed too and released openly.

Publications related to this project

• Talamini, Medvet, Nichele; Criticality-driven Evolution of Adaptable Morphologies of Voxel-Based Soft-Robots; Frontiers in Robotics and AI; 2021

• Nadizar, Medvet, Pellegrino, Zullich, Nichele; On the Effects of Pruning on Evolved Neural Controllers for Soft Robots; Workshop on Neuroevolution at Work (NEWK@GECCO); 2021

• Medvet, Pigozzi, Bartoli, Rochelli; Biodiversity in Evolved Voxel-based Soft Robots; ACM Genetic and Evolutionary Computation Conference (GECCO); 2021

• Ferigo, Iacca, Medvet; Beyond Body Shape and Brain: Evolving the Sensory Apparatus of Voxel-based Soft Robots; 24th European Conference on the Applications of Evolutionary Computation (EvoAPPS); 2021

• Medvet, Bartoli, De Lorenzo, Seriani, 2D-VSR-Sim: a Simulation Tool for the Optimization of 2-D Voxel-based Soft Robots, SoftwareX, 2020

• Medvet, Bartoli, De Lorenzo, Fidel, Evolution of Distributed Neural Controllers for Voxel-based Soft Robots, ACM Genetic and Evolutionary Computation Conference (GECCO), 2020, Cancun (Mexico)

• Medvet, Bartoli, De Lorenzo, Seriani, Design, Validation, and Case Studies of 2D-VSR-Sim, an Optimization-friendly Simulator of 2-D Voxel-based Soft Robots, arXiv, 2020

• Talamini, Medvet, Bartoli, De Lorenzo, Evolutionary Synthesis of Sensing Controllers for Voxel-based Soft Robots, Annual Conference on Artificial Life (ALIFE), 2019, Newcastle upon Tyne (United Kingdom)

Software related to this project

• 2D-VSR-Sim: https://github.com/ericmedvet/2dhmsr