Evolutionary robotics is a new and interesting area of research that uses Darwinian evolutionary principles to produce autonomous robots automatically. 

For the first time, a Frontiers in Robotics and AI publication shows that epigenetic factors affect embodied robot evolution. Scientists have used embodied robots to study epigenetics in robot evolution for the first time. This research's conceptual and practical method proves epigenetic components' importance.

In evolutionary robotics, a "gene pool" is artificially generated to yield genomes that encode different aspects of a robot's control system. The robots are then free to carry out the actions and tasks programmed into them by their "genetically" determined controllers; their fitness is measured by how successfully they do those duties. The robots are then permitted to reproduce by exchanging genetic material in a manner analogous to that of sexual reproduction in living organisms. However, development activities throughout an organism's life cause epigenetic alterations to the genome. The term "evo-devo" describes the relationship between evolution and development in biology and highlights the impact of environmental and developmental factors on phenotypic. 

Developmental robotics

Developmental robotics is related to areas like AI and ML, cognitive robotics and computational neuroscience because of its emphasis on adaptive intelligent robots. While it may use some of the methods developed in these areas, it is distinct from them in several important respects. Unlike traditional AI, it doesn't presume superior symbolic reasoning and instead emphasizes embodied and contextual sensorimotor and social skills instead of abstract symbolic problems. 

In contrast to cognitive robotics, it concerns the processes rather than the final products of cognitive development. Functional modelling of integrated architectures of development and learning is a crucial differentiator from computational neuroscience. In a broader sense, developmental robotics is distinguished by three main characteristics:

• It aims towards architectures and learning processes that are task-agnostic, meaning that the machine or robot can pick up tasks for which the engineer has no blueprint.
• It emphasizes the ability of an organism to learn new skills throughout its lifetime. It does not mean that one can learn "anything" or even "everything"; instead, it means that the horizons of one's knowledge can be expanded indefinitely in some but not all respects.
• Acquired expertise should progressively develop in complexity (while still being manageable).

Applications

Developmental robotics projects often focus on helping robots gain the same skills as human newborns due to the similar approach and technique. The acquisition of sensorimotor skills is one of the initial categories being studied. These include discovering one's body, structure and dynamics, hand-eye coordination, locomotion, object interaction, and tool use, with particular emphasis on finding and learning affordances. 

Developmental robots also teach simple social games like turn-taking, coordinated interaction, lexicons, syntax, and grammar, and they ground these linguistic skills into sensorimotor skills (called symbol grounding). The self/non-self distinction, attentional abilities, categorization systems, higher-level representations of affordances or social constructs, values, empathy, and theories of mind are being studied simultaneously.

Conclusion

Human and animal development research benefits from developmental robotics because researchers can use robots to test ideas and theories in a controlled environment.

Furthermore, evolutionary robotics is similar to but distinct from, developmental robotics. DevRob is concerned with how the structure of a single robot's control system evolves due to experience, whereas ER employs populations of robots that change through time.

Image source: Unsplash