Analyzing the Performance of Remote-Drivers on Transit Shuttle Short Routes
Using driving simulation, this project will study how a remote driver performs when operating a transit vehicle in complex scenarios that involve pedestrian-avoidance maneuvers. These types of maneuvers are typical at the University of Puerto Rico-Mayagüez campus due to the availability of satellite parking lots for students and the significant use of on-street parking by students. Transit shuttles traveling from the satellite parking locations constantly have to avoid pedestrians walking along the road. This type of scenario could be challenging to autonomous transit vehicles and can prompt the transfer of control to remote vehicle operators thus the focus of the project on the performance of a remote driver under such a scenario.
The type of maneuvers required to navigate the complex parking scenarios are often the result of localized behaviors which on its own present an interesting research challenge, e.g, understanding how the need for localized knowledge and understanding of cultural norms impact the feasibility of operating a vehicle remotely by drivers unfamiliar with the roadways. Driver performance, which will be primarily monitored through effective reaction time to events, will be studied using a driving simulator environment created in a game engine and which will be operated remotely using multiple vehicle views in a custom control environment. UW)
Current autonomous vehicle technology has been trained in traditional roadway conditions that do not necessarily represent the aggressive road user behavior and nuances of driving transit vehicles in highly congested environments with the combination of cars, pedestrians, and roadside parking. Therefore, until fully autonomous vehicles become an effective transit solution for an environment with highly localized behavior, remotely operated passenger vehicles that operate autonomously on most of the roadway can be used to manage a more extensive fleet of vehicles to satisfy the existing transit demand. The proposed research project will study the safety performance of a remotely operated passenger transit vehicle in a simulation scenario that replicates vehicle and pedestrian behavior typical to local highways in the University of Puerto Rico at Mayaguez (UPRM) campus that will be used as a test-bed in this project. Transfer of control issues aside, the remote operation of a vehicle warrants studying as it is not a simple task with deterministic safety impacts. Latency in the network that makes communication between a control station and a remotely operated vehicle possible could be detrimental to safety. Initially, data will be collected to document the most concerning situations related to a potential implementation of autonomous transit vehicles. After identifying critical safety conditions, scenarios will be developed to generate the simulation test-bed. Simulation experiments will be conducted using a simulation control interface created at the University of Wisconsin Madison. Experiments will include a group of drivers that are knowledgable of the shuttle route and another group that would not be familiar. Analysis of the reaction time data to events will be performed using the data collected in the Simulation Control Cabin to compare the performance of both groups of drivers. (UPR)