Luobo Kuaipao (Apollo Go) Tested by Politicians: It's About the Books, Not the Cars

Title: Luobo Kuaipao (Apollo Go) Experienced by Political Figures; First, Do the Operational Math

On September 12, during the 2026 CIFTIS, legislators and political figures from seven countries experienced Apollo Go. Romina Khurshid Alam, Pakistan's Coordinator for Climate Change and Environmental Affairs, said the ride was very smooth, like an experienced driver, and expressed hope to introduce it soon. One day, seven countries, one ride experience—these numbers seem light, but the signal behind them is heavy. Autonomous mobility services are moving from corporate launch events to public policy tables.

I discussed in a recent reply how embodied intelligence valuations will stratify. For warehouse robots, I care more about delivery costs and cash flow. Looking at Apollo Go now, the logic is similar. Passengers sit inside and evaluate smoothness, safety, and feeling like an old pro. What customers pay for is the system supporting these evaluations.

In warehouses, our biggest fear is mistaking a successful demo for a viable product. Anyone who has run robots in actual warehouses knows that getting a robot from point A to point B isn't hard; what's hard is when shelf positions are temporarily occupied by people, floors have reflections, forklifts suddenly reverse, or pallets tilt slightly and trigger alarms. Motion planning looks beautiful in simulation, but after an afternoon of on-site calibration, the map might still drift. I've just started working with loop closure detection these past few days, and the more I test, the more I understand: stable maps rely on sensors, calibration, and exception handling all providing a safety net.

Political figures experiencing Apollo Go is similar. Smoothness inside the car indicates that under the current road conditions, weather, and traffic flow, the system didn't have major issues. Cross-border deployment faces different road conditions. Different countries' speed limits, right-of-way rules, parking regulations, construction zones, mixed e-bike traffic, and pedestrian habits will all change prediction models. Has the deployment cost been calculated? Map updates, remote monitoring, local safety redundancy, accident liability, insurance, fleet maintenance—these aren't written into the experience reviews. Smoothness is just the entry ticket; business requires calculating costs.

The news headline mentions "green mobility as new quality productive forces." This expression is grand. For policymakers, green means emission reduction, energy, and traffic governance; for enterprises, green means power consumption, empty runs, charging, scheduling, and maintenance. Focusing only on concepts easily turns into propaganda. Implementation requires doing the math. To determine if an autonomous mobility service is truly green, you must at least look at whether vehicles are electrified, where the energy comes from, whether empty-run rates are suppressed by scheduling, whether remote O&M and data center energy consumption are accounted for, whether invalid driving caused by accident takeovers/detours is controllable, and whether it can integrate with public transport and slow-traffic systems in the target country.

I work on warehouse robots and often tell clients: don't ask how smart the robot is first; ask about unit picking costs, fault recovery time, and on-site engineer residency costs. In their warehouses, night-shift forklift drivers love taking shortcuts; if a robot misjudges once and stops the line, the client cares about the loss from the stoppage. Green mobility is the same. Political figures and legislators asking if their country can introduce it are essentially asking: Can this generate public benefits? Can it create new jobs? Can it reduce congestion and emissions? Can we avoid being dragged into public opinion crises due to technical accidents?

Apollo Go being experienced by legislators from seven countries shows that Chinese autonomous mobility has passed the "does it exist" stage and entered the "can it be replicated" stage. Replication is hardest regarding the system. The system includes regulatory interfaces. Can data cross borders? How do you get mapping survey licenses? How is accident liability defined? What is the ratio of remote safety operators? How does insurance compensate? The system includes local O&M. Who cleans the vehicles? Who calibrates sensors? Who handles temporary road closures? Who trains dispatchers? The system includes supply chain. Spare parts, charging, computing power, cloud scheduling, edge devices—price fluctuations affect gross margins.

This is a direct reminder for embodied intelligence entrepreneurs. Whether doing warehousing, delivery, inspection, or dual-arm manipulation, don't just stare at model capabilities. Clients want stable delivery. Team execution capability is reflected in exception handling. I agree with a judgment: scenarios with cash flow often don't look sexy enough. Warehouses, ports, industrial parks, hospital logistics—they don't make stories as easily as humanoid robots, but orders, payments, and maintenance contracts are clearer. Cash flow is harder than narrative; this holds true for autonomous driving going global too.

Public sector experiences are sometimes preludes to procurement. CIFTIS is more like a resource matching venue. When legislators from seven countries sit in the car, it changes some domestic perceptions of autonomous driving within those countries. For companies, this is a window to lower trust costs. A window doesn't equal an order. Orders depend on contract boundaries, acceptance criteria, after-sales SLAs, and failure rate compensation. When these countries actually invite autonomous mobility services in, the first questions political figures will ask are: Who is responsible if something goes wrong? How long until recovery? Orders ultimately come down to clauses on liability, recovery time, and compensation.