Developing food products from initial concepts to a validated prototype requires systematic exploration of formulations, processing conditions, and their interactions. For plant-based foods, this design space expands substantially as diverse protein sources, functional additives, and processing parameters create complexity that demands experimental throughput and consistency beyond manual laboratory workflows. This thesis advances automation in laboratory-scale plant-based food product development by integrating robotic systems, sensing methods, and optimization workflows.

For liquid food formulations, an automated experimentation platform was developed to enable repeatable processing and rheological characterization of plant-based beverages. Coupled with probabilistic surrogate modeling, the platform enables closed-loop, sample-efficient optimization of ingredient and process parameters to identify low-viscosity formulations. Extension to multi-objective optimization further enables systematic exploration of trade-offs between competing targets. For solid foods, an automated cooking platform integrates multimodal sensing during cooking with post-cooking mechanical characterization to enable repeatable assessment of plant-based meat alternatives. The resulting multimodal datasets allow quantitative comparison of cooked products and identify the sensing modalities most informative for distinguishing differences across formulations and food matrices. In the sensing domain, a robotic titration platform enables automated, computer-vision-based assessment of physical instability in liquid protein systems. Aggregation and sedimentation are quantified using Laplacian-based image analysis with a standard RGB camera, replacing subjective human visual inspection. At the laboratory infrastructure level, a developed Cartesian robot simplifies motion planning while supporting multi-robot collaboration, and concurrent operation of multiple instruments. Additionally, a large-language-model-based framework enables autonomous self-assessment of robotic laboratory readiness, autonomously detecting faults and unsafe conditions prior to experimentation.

Together, these contributions establish a foundation for scalable robotic experimentation in food laboratories, demonstrating how automation, sensing, and data-driven optimization can accelerate the development of plant-based food products.