Design of experiments meets immersive environment: Optimising eating atmosphere using artificial neural network

Area of Science:

• Cross-disciplinary application of statistical methods
• Food science and hospitality research
• Virtual reality and immersive environment studies

Background:

• Restaurant atmosphere significantly impacts perceived food quality and dining enjoyment.
• Virtual reality (VR) offers a cost-effective method for studying complex environmental scenarios.
• Design of Experiments (DOE) is a statistical approach for optimization, traditionally used in food development.

Purpose of the Study:

• To adapt DOE for optimizing restaurant atmosphere, specifically lighting and sound.
• To compare the predictive performance of Artificial Neural Network with Particle Swarm Optimization (ANN-PSO) against Response Surface Method (RSM).
• To determine optimal ambient conditions for enhancing situational appropriateness and food wanting in a virtual dining setting.

Main Methods:

• Utilized a Central Composite Design (CCD) within DOE to vary illumination and background sound levels.
• Employed an Artificial Neural Network with Particle Swarm Optimization (ANN-PSO) for modeling and optimization.
• Recruited 283 participants to rate two dishes in various virtual ambient conditions using immersive 360° videos.

Main Results:

• ANN-PSO model identified optimal lighting at 289 lux and -21.38 LUFS for virtual burger consumption.
• Optimal conditions for virtual pizza were found to be 186.9 lux and -30 LUFS.
• ANN-PSO demonstrated superior reliability and predictability compared to traditional RSM in this context.

Conclusions:

• The study successfully demonstrated the transferability of DOE methods to optimize virtual restaurant environments.
• Specific lighting and sound levels were identified to enhance virtual dining experiences for different food types.
• This research highlights the potential of integrating statistical optimization techniques with emerging technologies like VR in hospitality.