An interactive genetic algorithms system customizes robot appearance via cognitive noise filtering under embodied constraints

To address the inefficiency in evolutionary optimization caused by ambiguous user demands and cognitive noise in embodied robot appearance customization, this study proposes an Interactive Genetic Algorithms (IGAs) optimization framework integrating user preferences and embodied constraints. Firstly, a visual-semantic mapping mechanism was constructed to reduce initial cognitive noise through text-image-symbol associations. Secondly, an embodied constraint chromosome specification was established to intercept physically infeasible solutions using a knowledge graph. And finally, a progressive IGAs interaction model was optimized to guide user decision-making focus in stages. Compared to traditional IGAs, the system significantly improves customization efficiency: user evaluations decreased by 35.2%, user burden reduced by 30.4%, and iteration cycles shortened by 15%. The framework thereby enables low-cognitive-load interaction and establishes a user-driven design paradigm for highly constrained products, offering a methodological reference for the collaborative design of function-sensitive products.