KargoBot Launches Large-Scale AI Transport Network, Tests Cabless Heavy Trucks

The company came to Inner Mongolia five years ago to test platooning autonomous driving. It now operates an autonomous transport fleet in the hundred-vehicle range, with business extending from Inner Mongolia to energy transport regions in Xinjiang, Gansu, Shaanxi and Shanxi. It has also begun testing cabless transport robots. The route from Ordos westward to Qipanjing passes coal mines, power plants and industrial parks, where heavy trucks move large volumes of cargo along fixed routes; the area is one of KargoBot's main proving grounds for unmanned freight.

KargoBot's push for scale rests on route economics beginning to work. Since last year, the company has connected coal mines, factories and loading and unloading with Ordos Group and achieved positive unit economics at the route level in real transport tasks. With cabless transport robots in operation, KargoBot says per-vehicle economic benefits could rise from 20 percent to more than 30 percent.

CEO Wei Junqing summarized the traditional cost structure as '33211': tolls about 30 percent, energy about 30 percent, drivers about 20 percent, and maintenance and vehicle depreciation about 10 percent each. In recent years, new-energy heavy trucks have reduced some energy costs, but recruiting truck drivers has become harder and labor costs have continued to rise. Platooning changes the driver cost. Two trucks previously required two drivers; now one AI pilot can drive the lead truck and lead one unmanned truck. If one driver can reliably lead three or more vehicles in the future, labor costs would be diluted further. Based on current real operations, KargoBot says platooning can improve gross margin by about 10 percent to 18 percent compared with traditional manual transport after accounting for depreciation, maintenance and insurance.

A single profitable route is still far from scale. Moving from 10 vehicles to 100, and then toward 1,000, brings more operational problems. Bulk logistics has clear seasonal and cyclical swings, and cargo volumes on routes fluctuate. Predicting capacity demand and controlling dispatch, operations and maintenance costs become more important. KargoBot is therefore adding vehicles on mature routes, expanding from a dozen or so to the hundred-vehicle level, while copying validated models to new regions so that one line becomes several and lines form a network. The network has reached six provinces and involves 30 customers, eight automakers and about 30 ecosystem partners.

If KargoBot had to buy, operate and find cargo for every new route and every additional hundred vehicles itself, expansion would remain limited. The company plans to open mature routes to more participants. At present, unmanned transport is mainly self-operated or carried out with large logistics companies. Once a route's technology, product and operating plan are mature enough, small fleets of several to a dozen vehicles could also enter the network. KargoBot says deployment time for a single route has been shortened from eight to 12 months to nearly half that.

Platooning still presents operating challenges. KargoBot's platoons can have one pilot leading up to five following vehicles, forming a queue more than 100 meters long. In actual operations, it usually leads only one autonomous truck to avoid affecting other traffic. On public roads, autonomous heavy trucks still mix with passenger cars and other trucks. A car may suddenly cut between two platoon vehicles, or another heavy truck may cut in. The autonomous truck will sound its horn, slow down and increase the gap to leave space. Traffic lights can also separate the vehicles: if the lead truck passes but the rear truck is stopped, the lead waits while the rear continues using its own autonomous driving capability and rejoins later.

Platooning does not mean the rear vehicle simply imitates the lead. Because heavy trucks are tall, the rear vehicle's view is partly blocked. The lead truck's processed perception information is sent to the rear, along with throttle, braking and trajectory data. But the lead truck's information is an additional reference; the rear vehicle must retain independent judgment. If a vehicle cuts in or the lead's information fails, the rear vehicle still has to decelerate, avoid and decide its path on its own. Many routes also lack stable network signals. Some transport routes in Inner Mongolia and Xinjiang have weak or no signal, so communication between vehicles cannot rely on cellular networks; vehicles must be able to exchange information directly in no-signal areas. KargoBot therefore keeps an experienced driver in the lead truck to handle complex situations on open roads.

Platooning is not KargoBot's final product form. Since last year, some following vehicles have gone three to four months, or even five to six months, without human driving. KargoBot has now introduced cabless transport robots. Removing the cab increases usable transport space by about 25 percent and raises cargo capacity. In the future, following vehicles can be replaced by cabless robots. A fleet may therefore include human-driven heavy trucks, driverless trucks with cabs and fully cabless transport robots. Even if the lead truck eventually becomes driverless, platooning retains value. In planned bulk logistics, many vehicles already travel at the same time on the same routes, and multiple heavy trucks driving in formation can reduce wind resistance. KargoBot's tests show that a two-truck platoon can cut energy consumption by about 5 percent to 10 percent.

As drivers gradually leave the vehicle, the challenge for unmanned freight shifts from single-vehicle intelligence to the entire transport system. At scale, competition is no longer only about vehicle capability but about whether technology, vehicles, cargo, dispatch and operations can be organized into a stable system. For KargoBot, the next key step is to replicate its validated business model in more regions and scenarios.