About This Unity asset provides an end-to-end, Human-in-the-Loop (HITL) Multi-Objective Bayesian Optimization (MOBO) workflow built on botorch.org. It lets you declare design parameters and objectives in Unity, handles a Python backend for MOBO, and loops with users inside your Unity scene. The result is an efficient search over large design spaces, yielding trade-off designs on the Pareto front. Why this matters. Users typically have diverse preferences, needs, and abilities. Thus, manual design parameter tuning is often slow and potentially biased; A/B and grid search scale poorly. Instead, MOBO uses probabilistic surrogate models and principled acquisition to balance design exploration and exploitation, reducing the number of user trials required to reach a high-quality design for every user. Key Features Configure design parameters, objectives, and optimizer hyperparameters directly in Unity. Automatic, robust communication with a BoTorch-based MOBO process. Built-in integration with the QuestionnaireToolkit for explicit feedback in a HITL process; compatible with implicit telemetry. Automatic CSV logging of parameters, objectives, and hypervolume; warm-start from prior runs. Two example scenes demonstrating end-to-end optimization. Example Use Case To improve interface usability, treat selected UI attributes as design parameters x (e.g., button size, color contrast, spacing, animation duration) and optimize two objectives y: System Usability Scale (0–100, maximize) and task completion time (seconds, minimize). In each iteration t, the optimizer proposes a configuration xt; a participant completes a fixed task; Unity records time; the participant completes SUS; the posterior and acquisition function update; and the next xt+1 is selected. After several iterations, the system returns an estimated Pareto front containing Pareto-optimal interface designs that represent the best compromise between the design objectives. Release Highlights (v1.1.0) Click Game example scene Minimal 2-objective task (time-to-click, perceived difficulty). Ready to plug into the BO loop. Acquisition upgrade: qNEHVI → qLogNEHVI Replaced legacy qNoisyExpectedHypervolumeImprovement with qLogNoisyExpectedHypervolumeImprovement for better numerical stability. Same API. from botorch.acquisition.multi_objective.logei import qLogNoisyExpectedHypervolumeImprovement Simplified BO Manager inspector (Unity) Cleaner grouping, clearer hyperparameters, and safer defaults. Python bootstrap Auto-detects a Python executable and installs requirements.txt from StreamingAssets/BOData/Installation/ automatically. Sampling rule: 2d + 1 The number of sampling iterations now defaults to 2 * num_design_params + 1. Toggle manual override if needed. Upgrade notes Ensure StreamingAssets/BOData/Installation/requirements.txt is present; the runner installs packages automatically. If you previously set a fixed sampling count, enable "Set Sampling Iterations Manually" in the inspector to keep. Otherwise, the default is 2d+1. Changelog Add: Click Game example scene. Change: Switch to qLogNoisyExpectedHypervolumeImprovement. Improve: BO Manager inspector UX. Add: Default Python path detection + automatic requirements.txt install. Add: Automatic sampling iterations = 2d+1.