Reality-Assisted Evolution of Soft Robots through Large-Scale Physical Experimentation: A Review

Abstract

We introduce the framework of reality-assisted evolution to summarize a growing trend towards combining model-based and model-free approaches to improve the design of physically embodied soft robots. In silico, data-driven models build, adapt, and improve representations of the target system using real-world experimental data. By simulating huge numbers of virtual robots using these data-driven models, optimization algorithms can illuminate multiple design candidates for transference to the real world. In reality, large-scale physical experimentation facilitates the fabrication, testing, and analysis of multiple candidate designs. Automated assembly and reconfigurable modular systems enable significantly higher numbers of real-world design evaluations than previously possible. Large volumes of ground-truth data gathered via physical experimentation can be returned to the virtual environment to improve data-driven models and guide optimization. Grounding the design process in physical experimentation ensures that the complexity of virtual robot designs does not outpace the model limitations or available fabrication technologies. We outline key developments in the design of physically embodied soft robots in the framework of reality-assisted evolution.

From Bioinspiration to Computer Generation: Developments in Autonomous Soft Robot Design

“…The modeled design space can then be widely searched in simulation to illuminate high‐performing regions, with the best candidates physically verified and used to update the model. [ 150 ] Alternately, the experimental data can be used to refine a simulation environment by tuning its hyperparameters to minimize the reality gap and generate digital twins. In both cases, because the search is done in silico, a vastly larger region of the design space is exploitable compared with fully experimental methods.…”

Real-World Robot Evolution: Why Would it (not) Work?

“…Before finishing this section let me mention two related areas with an interesting future potential for evolving robots in new substrates. Soft robots go beyond the currently dominant mechatronical designs by using soft materials that can facilitate new types of sensing and actuation and these can be combined with evolutionary approaches Rieffel et al (2013); Laschi et al (2016); Howison et al (2021). However, to date such systems are either in simulation or are limited to "body parts".…”

Regenerating Soft Robots through Neural Cellular Automata

“…Recently, soft robots designed using computer simulations have recently been recreated in real robot using a variety of materials [19]. With the development of material science, a variety of soft robots that can change their shape have been born [12].…”