Autonomous Transit Alternatives

Last week German local public transit came to a stop while the union demanded better working conditions for transit operators. They asked for fewer working days and hours, to mitigate increased traffic stress and increase the attractiveness of transit jobs again. Operator availability and pay has put a limit on service recovery and expansion since the pandemic. I have written about automated metro trains and gondola lifts before, to reduce this dependency and offer higher frequency and longer operating hours. In this article I want to focus on automated shuttles and buses and how it may affect Metro's plans for the next few decades as recently discussed.

On-demand Shuttles

Elon Musk promised autonomous driving long ago, and investors poured billions into that opportunity. While the technology has made incredible progress on dedicated roads such as freeways, the uncertainties of the urban environment have proven to be more challenging, and the necessary technology more costly than anticipated. Many startups used up their funding and got acquired or closed. Rather than focusing on general driving (categorized as level 5), most companies have now focused on driving in well-defined urban areas (level 4) such as Beijing, Phoenix, San Francisco, Singapore, Dubai, or Hamburg. That way the AI can be trained to the local maps, topography, regulations, and challenges. They also added more sophisticated cameras and sensors, compute power, and AI than any of the current car manufacturers (Tesla, Mercedes, Volvo...) have included in their current cars. Waymo taxis can pull to the side if they identify an approaching emergency vehicle. However, there are still traffic situations the AI software cannot handle. If a car gets stuck, a driver can be called for manual rescue. Vay is proposing to use their car's sophisticated cameras to allow a human operator to take control of the taxi remotely instead. Companies such as Baidu, Zoox, Waymo and Cruise started to offer mobile app based rideshare (robotaxi) services to compete with Uber or Lyft but challenges remain. After blocking intersections and rail tracks, evading police, and accidents, Cruise paused driverless operation recently. Snow and even rain can create insurmountable challenges.

In the meantime, Metro is experimenting with on-demand shuttles with their Metro Flex offering (and Bellevue with Bellhop). Riders can use a phone app to request a trip ahead of time, and the app provides an estimated pickup time and location nearby. The Metro Flex app and service are provided by Via, which operates the minivans on Metro's behalf. It serves areas in South Seattle, Issaquah, Kent and Juanita, currently underserved by transit routes. The minivans are operated by regular drivers, but Via in partnership with May Mobility and Moia (Volkswagen) are already testing robotaxis for this purpose. Rideshare vehicles may reduce parking needs, but they can clog up tight neighborhoods just as badly as any other vehicle. If a robotaxi gets confused or in an accident, it may require calling a human driver to the scene. In the meantime the taxi may block traffic, even emergency vehicles. This has already happened in San Francisco. Even trams got blocked when taxis blocked tracks.

Uber and Lyft had tried to increase scale by pooling multiple trips into a combined trip. They dropped this effort when the detours created undue delays. Via and Moia are asking riders to request their ride ahead of time so that their software can pool multiple rides along a corridor into a single ride. Instead of picking you up at home, the app instructs you to walk to a virtual bus stop" close by where it is easier and faster for a shuttle to pick you up. That way the shuttle may also avoid getting stuck in tight little streets with less mapping data and more obstacles. Instead of using small cars or a minivan, they are also considering using slightly larger vans. As long as these shuttles only serve destinations within a small neighborhood, they might become quite useful. If the area also includes a frequent transit connection, this may become a great solution to extend transit services to suburban neighborhoods. Without having to pay drivers, you could run enough of these shuttles to offer a viable alternative to providing parking garages next to a transit station. Adoption would certainly increase if fares would be integrated with regular transit and the app would be integrated into regular transit apps provided by transit agencies or platforms like Google or Apple Maps.

Autonomous Buses

Though buses may encounter similar challenges as shuttles in smaller neighborhoods, in an urban environment bus routes often operate on major arteries, separate bus lanes, or even freeways such as our ST Express or soon Stride buses. The more dedicated and separated the right-of-way, the less challenges for autonomous driving. Companies such as Aurora are already testing autonomous trucks along freeway routes. Right before their destination they meet a driver which operates the truck to its final urban destination. A Stride bus may run autonomously along the freeway and stop at freeway stations. An Express bus may go from downtown to Lynnwood autonomously where it is met by a driver who drives the bus through the adjacent neighborhoods to Mill Creek.

While fully autonomous driving is still being perfected, some agencies are also experimenting with platooning, where a traditionally operated bus is followed by one or more driver-less buses in a convoy. Platooning is being tested in Munich as a more flexible alternative to articulated buses, and in New Jersey to get more buses through the Lincoln tunnel. West Seattle buses (C, H, 21) could form a platoon before the West Seattle bridge and ride with a single driver along Highway 99 to stations downtown and back to West Seattle, where they meet their drivers again to fan out across West Seattle. On a freeway this may work well, but there are limits to the length of a platoon in an urban environment because stations need to provide enough space and signal priority may be a requirement, so that a platoon can safely cross an intersection without the signal turning between buses.

Currently buses have more route flexibility but higher operating costs per rider than trains because you need more operators and more vehicles to maintain. As electric buses become more standardized, I expect that maintenance cost will be lower as electric motors need less care than a diesel hybrid (neither combustion engine nor gearbox, no oil change) and they last longer. Battery buses are not quite as efficient as trolley buses and we will need more as they will need to stop for charging. Once you have electric motors (powered by battery or overhead) and driver-less operation, bus operation becomes more competitive to automated trains. The main remaining difference is the right-of-way. While buses can share the road, the more interactions the more you need to anticipate disruptions and delays. With dedicated bus lanes and signal priority, speed and reliability increases. With a dedicated busway it reaches similar efficiencies as an automated train. Trains may still provide a bit higher ride quality while buses have more route flexibility.

Conclusion

While I believe autonomous shuttles and buses are on the horizon, it has taken longer than anticipated and nobody knows when they will become generally available. In the meantime we will see some more controlled applications. Once autonomous vehicles, become mainstream our transit network may change. We might see more high frequency/capacity transit along major arteries or freeways with fewer stops and the use of shuttles to get people back and forth from/to those stations. Shuttles may serve the suburbs and replace the current coverage lines" which currently serve less dense neighborhoods but often operate infrequently. Therefore, Metro's network in 2050 may look quite different from the lines anticipated by Metro Connects.