Evolutionary Robotics and Morphological Design

Evolutionary robotics unites principles of natural evolution with robotic engineering to generate novel body plans and control systems through iterative selection, variation and heredity. Rather than designing physical form and neural control separately, this field treats the robot as a single evolving entity, optimising morphology and behaviour in tandem. Morphological design explores how shape, material distribution and modular architecture influence a robot’s capacity for locomotion, manipulation and environmental interaction. Advances in real‐world fabrication and model‐free evaluation have enabled physical instantiation of evolved designs, narrowing the reality gap between simulation and practice. Computational frameworks now combine deep learning, evolutionary algorithms and diversity‐oriented search to foster ‘morphological intelligence’, where evolved bodies inherently facilitate the learning of new tasks. Such integrated evolution of form and function holds promise for adaptive systems in search and rescue, planetary exploration and soft robotics, offering robustness and versatility in unstructured and dynamic environments.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

Quality‐diversity algorithms have been shown to excel at simultaneously producing high‐performance and morphologically diverse robot designs. In modular robotics, a framework based on MAP‐Elites has demonstrated the generation of stepping‐stone morphologies that not only achieve superior locomotion speed but also enable rapid re‐optimisation when transferred to novel terrains. This approach highlights the importance of diversity preservation for long‐term evolvability and adaptability. In parallel, fully autonomous hardware platforms have been developed that fabricate, assemble and test evolved morphologies in the real world. One such system employs easily assembled magnetic modules and visual fiducial tracking to deploy up to ten different evolved body plans in under five minutes. The close match between simulated and physical performance in these experiments underscores the maturity of end‐to‐end evolutionary loops for morphological design in practical settings.

Evolutionary Robotics and Morphological Design publication trend

The graph below shows the total number of articles in evolutionary robotics and morphological design across all publications each year (not limited to Nature Index journals).

Technical terms

Evolutionary Robotics: An approach that applies evolutionary algorithms to co‐optimise robot morphologies and control policies as integrated systems.

Morphological Design: The process of shaping robot bodies—through computational evolution or modular assembly—to enhance adaptability and performance.

Quality Diversity: A search paradigm that seeks both high-performing solutions and a wide variety of behaviours or form factors within a single run.

Embodied Intelligence: The concept that intelligent behaviour arises from the tight coupling between a robot’s physical form and its environment.

Modular Robotics: A design strategy using interchangeable units that can be reconfigured to produce diverse morphologies and functional specialisations.

References

1. Evolutionary Robotics: What, Why, and Where to. Frontiers in Robotics and AI (2015).
2. Embodied intelligence via learning and evolution. Nature Communications (2021).
3. Morphological Evolution of Physical Robots through Model-Free Phenotype Development. PLOS ONE (2015).
4. MAP-Elites Enables Powerful Stepping Stones and Diversity for Modular Robotics. Frontiers in Robotics and AI (2021).
5. EMERGE Modular Robot: A Tool for Fast Deployment of Evolved Robots. Frontiers in Robotics and AI (2021).