Simulation of Autonomous RoboShuttles in Shared Space

Abstract

In the mobility sector, autonomous driving will become more and more part of our daily life. Most of all, in public transportation the research to exploit the new possibilities of autonomous driving has increased drastically. But the problem of the last mile is still unsolved, for example. The last mile is the problem to transport people from a transportation network (examples of endpoints are parking lots or bus station) to their final destination. A promising solution to this problem are autonomous RoboShuttles. Because of their low velocity, they can operate in the shared space where pedestrians and vehicles share the same traffic area. Therefore, in a shared space the interaction between them is greater than under normal traffic conditions, where the traffic flows are separated as much as possible. Through this higher interaction, new requirements on the autonomous vehicle arise. To explore the new requirements and to understand the interaction of a RoboShuttle in a shared space, a simulation scenario in SUMO is set up.

SUMO4AV: An Environment to Simulate Scenarios for Shared Autonomous Vehicle Fleets with SUMO Based on OpenStreetMap Data

“…The second difference is that scenarios are abstracted by taking the strengths of the SUMO simulator and combining them with those of OpenAI Gym. The work of Gasper et al [7] deals with the use of autonomous shuttles. Here, the autonomous shuttles move autonomously in an area shared with pedestrians.…”

“…Then an AV needs to be dispatched to this customer. SUMO provides a configurable Taxi dispatch algorithm to simulate this case of demand responsive transport (DRT) 7 . All strategies described in Table 1 use the SUMO Taxi dispatch algorithm, but they differ in the way how the requests are processed.…”