StartLux: Assembly-Line Replenishment and Yielding Simulation

One blue mobile robot carries two bins containing eight boards each. Three assembly lines each request eight boards, so this trip cannot serve every order. The decision problem combines limited cargo, line priority, receiving conditions and a shared traffic crossing. The video follows a complete trip: choose line A, slow and yield while a yellow transport vehicle crosses, resume delivery, push the first bin onto A's receiving table, then deliver the second bin to B and withdraw empty to a waiting point. The final screen reports 16 of 16 boards delivered. Line C remains for a later trip; this recording does not show a mid-trip reassignment. The supplied notes describe StartLux receiving structured orders, shortages, deadlines, available cargo, robot state and route/traffic costs calculated by the harness. Selecting a delivery is a compound skill, rather than an instruction for individual wheel movements. Repeated choices to continue toward B preserve an existing task; they are not extra deliveries. The harness drives the differential-wheel robot, shapes speed, enforces traffic protection, aligns it at the receiving table and pushes each bin off using a separate actuator. Delivery is counted only after the bin is supported and stable on the table. The documentation also distinguishes the initial protective stop from the model's later explicit yield choice: the earliest braking is not attributed to StartLux. Evidence and limits: the listing is based on the supplied MuJoCo recording and implementation notes, without rerunning the simulation or independently checking the original API trace. The blue AMR is the only model-directed vehicle; yellow vehicles are scheduled background traffic. This demonstrates one simulated replenishment trip, not a validated factory fleet or safety certification. On-screen probabilities and timings are not delivery-success rates or an independent latency benchmark. The preserved video uses Chinese interface text and has no audio.