Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TRAVERSE: Traffic-Responsive Autonomous Vehicle Experience & Rare-event Simulation for Enhanced safety

Sandeep Thalapanane, Sandip Sharan Senthil Kumar, Guru Nandhan Appiya Dilipkumar Peethambari, Sourang SriHari, Laura Zheng, Julio Poveda, Ming C. Lin

TL;DR

The paper addresses the lack of adversarial scenario data for autonomous driving by introducing TRAVERSE, a portable VR driving platform that emulates rare, pre-crash events. It integrates RoadRunner, SUMO, and Unity with consumer hardware (Meta Quest Pro and Logitech wheel) to produce realistic, customizable pre-crash scenarios based on NHTSA typology. The authors demonstrate both qualitative and quantitative benefits over existing simulators, including improved immersion, realism, and data capture capabilities, while acknowledging trade-offs in FPS at higher traffic densities. The work highlights practical implications for safer autonomous systems and outlines future directions toward differentiable physics and driver-vehicle interaction modeling for enhanced learning and trajectory prediction.

Abstract

Data for training learning-enabled self-driving cars in the physical world are typically collected in a safe, normal environment. Such data distribution often engenders a strong bias towards safe driving, making self-driving cars unprepared when encountering adversarial scenarios like unexpected accidents. Due to a dearth of such adverse data that is unrealistic for drivers to collect, autonomous vehicles can perform poorly when experiencing such rare events. This work addresses much-needed research by having participants drive a VR vehicle simulator going through simulated traffic with various types of accidental scenarios. It aims to understand human responses and behaviors in simulated accidents, contributing to our understanding of driving dynamics and safety. The simulation framework adopts a robust traffic simulation and is rendered using the Unity Game Engine. Furthermore, the simulation framework is built with portable, light-weight immersive driving simulator hardware, lowering the resource barrier for studies in autonomous driving research. Keywords: Rare Events, Traffic Simulation, Autonomous Driving, Virtual Reality, User Studies

TRAVERSE: Traffic-Responsive Autonomous Vehicle Experience & Rare-event Simulation for Enhanced safety

TL;DR

The paper addresses the lack of adversarial scenario data for autonomous driving by introducing TRAVERSE, a portable VR driving platform that emulates rare, pre-crash events. It integrates RoadRunner, SUMO, and Unity with consumer hardware (Meta Quest Pro and Logitech wheel) to produce realistic, customizable pre-crash scenarios based on NHTSA typology. The authors demonstrate both qualitative and quantitative benefits over existing simulators, including improved immersion, realism, and data capture capabilities, while acknowledging trade-offs in FPS at higher traffic densities. The work highlights practical implications for safer autonomous systems and outlines future directions toward differentiable physics and driver-vehicle interaction modeling for enhanced learning and trajectory prediction.

Abstract

Data for training learning-enabled self-driving cars in the physical world are typically collected in a safe, normal environment. Such data distribution often engenders a strong bias towards safe driving, making self-driving cars unprepared when encountering adversarial scenarios like unexpected accidents. Due to a dearth of such adverse data that is unrealistic for drivers to collect, autonomous vehicles can perform poorly when experiencing such rare events. This work addresses much-needed research by having participants drive a VR vehicle simulator going through simulated traffic with various types of accidental scenarios. It aims to understand human responses and behaviors in simulated accidents, contributing to our understanding of driving dynamics and safety. The simulation framework adopts a robust traffic simulation and is rendered using the Unity Game Engine. Furthermore, the simulation framework is built with portable, light-weight immersive driving simulator hardware, lowering the resource barrier for studies in autonomous driving research. Keywords: Rare Events, Traffic Simulation, Autonomous Driving, Virtual Reality, User Studies
Paper Structure (17 sections, 9 figures, 1 table)

This paper contains 17 sections, 9 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Works
  3. High Fidelity Driving Simulators
  4. Low Fidelity Driving Simulators
  5. VR Driving Simulators
  6. Driving Simulator
  7. Simulator Requirements
  8. Design Implementation
  9. Hardware Components
  10. Softwares
  11. Data Collection
  12. Scenario-Based Assessments
  13. User Study Evaluation on Simulation Fidelity
  14. Results
  15. Qualitative Comparison To Existing Simulators
  16. ...and 2 more sections

Figures (9)

  • Figure 1: Users engage in VR driving simulation using Meta Quest Pro VR and Logitech steering wheel, with the user's view displayed on the monitor and the HMD.
  • Figure 2: Simulator System Diagram. A schematic representation depicting the integrated flow of the simulator, encompassing the three software components: Road Network generation, Traffic simulation, and 3D simulation, aimed at enhancing the user study experience. This framework provides a comprehensive environment for simulating and studying driving behaviors and scenarios
  • Figure 3: NHTSA pre-crash scenarios implemented in our virtual reality driving simulation. Scenarios include Sudden Lane Change Interaction, T-Bone Crash, Sudden Vehicle Stop in Front, Vehicle Running Red Lights, Sudden Deer Crossing, Crash at Roundabout, Crash in Ramp Merger, and Jaywalking Pedestrian Crash (from left to right, top to bottom). These scenarios were designed to simulate various challenging driving situations mentioned in NHTSA pre-crash typology najm2007definition to enhance the realism and effectiveness of the simulation
  • Figure 4: First-person and Third-person perspectives. (from left to right) of the simulator featuring rear-view and side-view mirrors, a speedometer, and driving modes (D/R) in a city environment with simulated traffic vehicles, and pedestrians. This view provides a realistic representation of the driving experience, enhancing the immersion for users
  • Figure 5: User study design for systematic evaluation of VR driving simulation quality. The user study is evaluated in two stages (green and orange arrows). Each evaluated participant will complete simulator sickness evaluations before and after each stage, in addition to a simulator evaluation questionnaire (black arrows). In both stages, the participant drives through NHTSA pre-crash scenarios of each simulator and evaluates the quality in several dimensions, further detailed in Section \ref{['sec:user_study_description']}. The order at which the user experiences each simulator is randomized during the study to account for order bias. Additionally, users complete each stage on two separate days, in order to minimize sickness effects carrying over to the second stage.
  • ...and 4 more figures