Vision · Language · Action
Given multimodal observations, language instructions, and optional state-action histories, the engine generates episode-, step-, and navigation-level reasoning traces together with executable targets, followed by automatic parsing and semi-automatic quality verification.
Multimodal observations and language instructions are fused for structured reasoning, which conditions a task-level action decoder to predict locomotion (Vx, Vy, ω) and a discrete task-level behavior primitive α for mobile robot control.
For each multimodal input, the policy samples multiple structured outputs whose induced actions are evaluated by movement, behavior, and format rewards. Group-relative advantages, together with KL regularization to a frozen reference policy, are then used for offline policy optimization.