Emergent Design of Dynamical Behavior

Abstract

As a mechanical system has larger number of degree-of freedom (DOF) of motion in order to generate diverse behavior, it becomes more difficult to design the control scheme for its motion to achieve a specific function required for the system. This paper proposes a novel design methodology for the control scheme in order to deal with the above-mentioned difficulty. The design methodology proposed is inspired from the concept of emergence observed in living organisms, that is, diverse behavior is emerged to an ultra multi-DOF dynamic system only with a single set of local rules using local information between neighboring subsystems. This study simplifies the ultra multi-DOF system as an 11×11 lattice mass system, neighboring masses are connected with a linear actuator, and aims to prevent a target mass point (control subsystem) from moving even if any external forces are input to the system, which we call this function “displacement controllability.” Hence the P/D gains of the actuators are turned ON/OFF by the proposed local rule for displacement controllability. As a result of computation, the displacement controllability was successfully emerged to the mass system under the various environments, i.e., diverse external force inputs. That means, the diverse behavior dealing with various environments was generated to the system only using the single local rule. Consequently, the effectiveness of the proposed behavior design methodology for an ultra-multi-DOF system was confirmed.