Optimizing Departures of Automated Vehicles From Highways While Maintaining Mainline Capacity

Automated vehicles have the potential to revolutionize the nation's transportation systems as they promise to dramatically reduce congestion, accidents, and fuel usage. Namely, as it becomes possible to precisely exert control on and coordinate the movement and placement of vehicles along a stretch of highway, the separation distance between vehicles can be reduced, thus increasing flow and minimizing congestion. Precise and coordinated vehicle controls and placements also improve predictability, resulting in fewer instances of sudden braking, which reduce fuel usage and accidents. Most existing research has focused on the steady-state behaviors and operations of automated vehicles, such as platooning, and assumes complete knowledge of the system, e.g., the number of vehicles and their destinations and/or neglects dynamic or transition operations such as exiting a highway and lane changing. Uncoordinated lane-changing and exiting behaviors by automated vehicles can significantly reduce the flow of traffic as vehicles will require larger separations, are forced to slow down, or worse, collide. In this paper, the authors present a collision-free runtime approach to efficiently organize the departures of automated vehicles from a highway environment while maintaining highway capacity in extremely dynamic conditions. To maximize the number of safe departures, the key ideas are to: 1) determine when, and where to, an exiting vehicle should lane change in order to make a successful exit given current traffic conditions as provided by connected vehicle technology and 2) execute the actual lane-change operations using a reservation-based approach. Simulation results show that, by coordinating vehicles' behaviors, traffic flow can be improved by up to five times of today's typical flow while ensuring a 100% exit success rate in a collision-free manner.

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01619109
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Dec 21 2016 11:29AM