Evolutionary Soft Robotics
Age-Fitness pareto optimization for designing the morphology of soft bodied creature
Morphological evolution in evolutionary robotics is an important approach for discovering new robot systems and developmental histories of many creatures. Conventional evolutionary robotics used a rigid-bodied robot by human design in computer simulation to verify the effective robotic control systems. However, it is hard to apply the findings of the conventional evolutionary robotics approaches into the development of a real robot like a creature from the evolved robot because there is major difference between the dynamics of the virtual robot and the real robot. Further, we potentially expect that the developed robot controls a desirable behavior from the robot morphology through any experiences.
In this research project, we propose a new method for evolving more complex and controllable soft-bodied robots without some developmental biases in order to get some innovative robotic findings by using the evolved virtual robot. This work found that the evolved soft-bodied robots have more versatility than most of conventional rigid-bodied robots and they got a lot of unintentional morphologies and behavior control systems like real creatures.