Evocation of Behaviors Based on Affordances in Industrial Environments as Compared to Biological Cases

Abstract

The concept of affordance is introduced to generate robot behaviors for performing tasks. In disassembly of a mechanical product, the disassembly process is considered to proceed as the product offers the affordances for behaviors of removing its parts and the agent, may it be a human or a robot, accepts them. Considering analogically that the agents are of a biological species, if they are placed in the environment rich with disassembly objects and the success of disassembly leads to the preservation of the species, the agents will acquire more efficient behaviors. An agent may find plural affordances in the environment simultaneously and the problem arises about which affordance to accept. We introduce regulations in accepting affordances, which correspond to the known psychological phenomena of adaptation and early selection. We claim that these regulations are a new kind of affordances and that the fundamental and regulative affordances together constitute an architecture for the integral behavior of the agent. The simulation using the Genetic Algorithm demonstrates that the disassembly robots evolve themselves for higher efficiency by developing regulations in accepting affordances in the environment with specific products.