Morphological change in machines accelerates the evolution of robust behavior.

Area of Science:

• Robotics
• Evolutionary Computation
• Developmental Biology

Background:

• Animals undergo significant lifelong neurological and morphological changes.
• Current robots lack the ability to grow new structures while operating, limiting their adaptability.
• The benefit of morphological change for machine behavior acquisition is not well understood.

Purpose of the Study:

• To investigate if morphological change during a robot's lifetime aids in the rapid evolution of robust behaviors.
• To explore the role of the evolution of development in machine learning and behavior generation.

Main Methods:

• Evolving populations of simulated robots with a developmental process.
• Comparing populations where robots transition from anguilliform to legged forms versus those that only evolve legged forms.
• Analyzing the speed and robustness of gait evolution in the final legged robot morphology.

Main Results:

• Robots transitioning through anguilliform to legged forms evolved gaits more rapidly.
• The evolved gaits for the legged form were more robust in the transitional group.
• This suggests that developmental processes can enhance evolutionary outcomes.

Conclusions:

• Morphological change and the evolution of development are crucial for automatically generating robust machine behaviors.
• This study provides a framework for studying the evolution of development and its link to robust behavior in biological systems.
• Incorporating developmental processes can accelerate and improve the robustness of evolved robot behaviors.