How flavor control works
FlavoRotor does not directly command taste. It controls and records the growing environment, measures how a specific plant responds, and learns crop-specific recipes that reproduce a bounded chemical and sensory outcome. A recipe becomes valid only after calibrated execution and replication.
Control chain
Nine sequential gates from target to versioned recipe release.
- 01target sensory profile
- 02crop and cultivar selection
- 03measured cultivation recipe
- 04calibrated actuator execution
- 05verified plant-level environment
- 06chemical and physical measurement
- 07appropriate blinded sensory measurement
- 08independent replication
- 09versioned recipe release
Controlled inputs
Seven cultivation inputs that FlavoRotor controls and measures independently.
- 01light spectrum, intensity, photoperiod and DLI
- 02nutrient composition and nitrogen-potassium balance
- 03bulk EC and controlled salinity range
- 04pH management
- 05solution temperature and root-zone conditions
- 06rotation and periodic root-zone exposure
- 07harvest timing and post-harvest protocol
Crop-specific targets
Each crop has a defined set of inputs to test and primary outcomes to measure.
basil
A named basil cultivar with a stronger bounded descriptor profile and a defined volatile pattern.
Inputs to test
- spectral composition
- DLI
- mechanical treatment
- harvest stage
Primary outcomes
- selected aroma volatiles
- descriptive aroma
- growth and morphology
arugula
A named arugula cultivar with a reproducible pungency range without unacceptable yield or nitrate trade-offs.
Inputs to test
- EC range
- nutrient composition
- light
- harvest stage
Primary outcomes
- glucosinolate-related phytochemicals
- pungency
- bitterness
- yield
- nitrate accumulation
lettuce
A named lettuce cultivar with a bounded bitterness and texture profile at an acceptable yield.
Inputs to test
- nitrogen regime
- light schedule
- pH
- EC
- root-zone temperature
Primary outcomes
- bitterness
- texture
- quality
- physiological response
mint
A named mint species and cultivar with a reproducible essential-oil and sensory profile.
Inputs to test
- controlled salinity range
- light
- nutrient composition
- harvest stage
Primary outcomes
- essential-oil percentage
- essential-oil composition
- menthol-related descriptors
- descriptive aroma
strawberry
A named strawberry cultivar with a bounded sweetness-acidity-aroma profile and defined firmness.
Inputs to test
- nitrogen-potassium relationship
- nutrient formulation
- light
- root-zone system
- harvest maturity
Primary outcomes
- soluble solids
- titratable acidity
- sugar-acid balance
- volatile profile
- firmness
- sensory response
Recipe prediction model
The model is fitted from FlavoRotor data after baseline, single-factor and interaction experiments. It is never seeded with invented nutrient-to-flavour coefficients.
- ŷ
- predicted crop-specific outcome
- x
- measured cultivation vector
- g
- genotype, including crop and cultivar
- s
- system state and calibrated hardware condition
- t
- developmental stage and timing
- ε
- unexplained variation
The output is a bounded prediction with uncertainty for a named crop, cultivar, system version and protocol, not a universal flavour slider.
Recipe release gate
A recipe is versioned and published only after all conditions are met.
- crop, cultivar and biological material are identified
- machine inputs are calibrated
- the actual environment is measured
- the protocol is frozen before outcome analysis
- chemical or physical endpoints match the claim
- the sensory method matches the claim
- uncertainty and effect size are reported
- an independent replication succeeds
- the recipe receives a version and declared validity range