Deploying a full-size SAE Level-4 automated bus in Michigan’s Sleeping Bear Dunes, overcoming GNSS-denied conditions with ADASTEC’s flowride.ai.
This past summer, we took on a project unlike any we had done before: bringing a full-size, electric, Level-4 automated bus to one of Michigan’s most remote and visited natural destinations, Sleeping Bear Dunes National Lakeshore. No internet connection, no GNSS/GPS reception, no previous public transit services, steep roads, and extremely remote! Exactly where you would need a real transit service.
Between August 22 and September 29 of 2024, the vehicle operated on scenic yet technically demanding routes. The automated bus ran on two loops, a 4.3-mile short route and a 7.4-mile long route, completing over 130 trips during the six-week period. Throughout the route, the breathtaking views came with their own set of challenges: the park’s terrain included narrow roads with limited passing space, variable lighting conditions under dense tree canopies, and steep drop-offs along certain sections, all of which required precise navigation and robust safety measures. It operated fully automated, covering more than 600 miles throughout the deployment. Despite the park’s limited connectivity and steep gradients, the system successfully met all operational targets for automation, safety, and reliability.
Seasonal tourism patterns meant that traffic volumes could change rapidly within a single day, creating unpredictable operating conditions. The route network demanded the ability to accommodate both paved and unpaved segments and to safely navigate around pedestrians, bicyclists, and wildlife in a remote environment.
Operations needed to remain reliable in varying summer weather conditions and during changing light levels under dense tree cover, even in areas with little or no satellite coverage.
That’s where flowride.ai, ADASTEC’s automated driving software platform, came in. The vehicle was able to operate safely in these complex and constantly changing conditions. Despite these challenges, the system consistently delivered reliable service, carrying passengers and proving that a self contained autonomy platform can thrive even when completely cut off from the cloud.
This deployment was not only a public demonstration but also part of an academic study titled “Operational Resilience and Challenges from Full-Sized SAE Level-4 Automated Bus Deployment in a National Park Environment” presented at the IEEE SOLI 2025 Conference. The research, co-authored by ADASTEC engineers, analyzed system resilience in GNSS-denied environments, passenger experience, and weather performance metrics. The study provides one of the first real-world deployments and operational insights of a full-size SAE Level-4 automated bus operating in mixed traffic under remote and GNSS-limited conditions.
Rural and infrastructure-limited environments demand autonomy that is independent of external networks. This rapid deployment underscored the value of onboard decision making and vehicle designs that anticipate environmental unpredictability.
This experience reinforces how automated transit, especially when it can operate without reliance on external networks, can be a viable solution not only for remote parks but also for high cost urban and suburban regions facing service cutbacks.
Findings from the Sleeping Bear Dunes project revealed that fully self-contained autonomy stacks can sustain extended operation without cloud connectivity or continuous GNSS support. This proves a scalable model for national parks, rural communities, and disaster recovery zones where traditional infrastructure is limited. These results are already guiding follow-up deployments and research collaborations not only across Michigan and other U.S. states but also within ADASTEC’s presence in Europe and beyond.
This rapid deployment experience at Sleeping Bear Dunes is shaping how we approach future deployments, from large university campuses to remote corridors. The insights gained here will help guide collaborations with OEMs, suppliers, and transit agencies as we work toward reliable, inclusive, and accessible automated mobility solutions.
We are grateful to be in collaboration with the Michigan Department of Transportation (MDOT), the Office of Future Mobility and Electrification (OFME), the National Park Service Washington Office (WASO), the National Park Service Sleeping Bear Dunes National Lakeshore (SLBE), U.S. DOT Volpe Center, NextEnergy, ARIBO, and all who contributed to making this rapid deployment a success. This is one step closer to realizing the full potential of SAE Level-4 automated public transit.
Special thanks to Selen Kandoğan, Business Development Intern at ADASTEC, for her contributions to this article and for highlighting the key outcomes of the Sleeping Bear Dunes deployment.
