Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Vehicle System for Navigating Among Vulnerable Road Users Including Remote Operation

Oscar de Groot, Alberto Bertipaglia, Hidde Boekema, Vishrut Jain, Marcell Kegl, Varun Kotian, Ted Lentsch, Yancong Lin, Chrysovalanto Messiou, Emma Schippers, Farzam Tajdari, Shiming Wang, Zimin Xia, Mubariz Zaffar, Ronald Ensing, Mario Garzon, Javier Alonso-Mora, Holger Caesar, Laura Ferranti, Riender Happee, Julian F. P. Kooij, Georgios Papaioannou, Barys Shyrokau, Dariu M. Gavrila

TL;DR

The paper tackles safe navigation around VRUs in low-speed, cluttered environments by introducing a topology-driven Model Predictive Control (T-MPC) motion planner that generates multiple probabilistic, parallel trajectories and accommodates non-passing strategies. A learned encoder–decoder maps guidance trajectories to optimized local plans under a chance-constrained MPC framework, and a Remote Control Tower provides a human-in-the-loop fallback for edge cases. The system is demonstrated on a Toyota Prius prototype with a full perception-localization-planning-control stack and shows superior safety and efficiency in simulations against baselines, as well as real-world tests on a closed track in both autonomous and remote operation modes. This work advances practical VRU navigation by combining topology-based trajectory optimization with remote operation, enabling safer operation in unstructured urban contexts and offering a structured pathway for handling exceptional situations through human oversight.

Abstract

We present a vehicle system capable of navigating safely and efficiently around Vulnerable Road Users (VRUs), such as pedestrians and cyclists. The system comprises key modules for environment perception, localization and mapping, motion planning, and control, integrated into a prototype vehicle. A key innovation is a motion planner based on Topology-driven Model Predictive Control (T-MPC). The guidance layer generates multiple trajectories in parallel, each representing a distinct strategy for obstacle avoidance or non-passing. The underlying trajectory optimization constrains the joint probability of collision with VRUs under generic uncertainties. To address extraordinary situations ("edge cases") that go beyond the autonomous capabilities - such as construction zones or encounters with emergency responders - the system includes an option for remote human operation, supported by visual and haptic guidance. In simulation, our motion planner outperforms three baseline approaches in terms of safety and efficiency. We also demonstrate the full system in prototype vehicle tests on a closed track, both in autonomous and remotely operated modes.

A Vehicle System for Navigating Among Vulnerable Road Users Including Remote Operation

TL;DR

The paper tackles safe navigation around VRUs in low-speed, cluttered environments by introducing a topology-driven Model Predictive Control (T-MPC) motion planner that generates multiple probabilistic, parallel trajectories and accommodates non-passing strategies. A learned encoder–decoder maps guidance trajectories to optimized local plans under a chance-constrained MPC framework, and a Remote Control Tower provides a human-in-the-loop fallback for edge cases. The system is demonstrated on a Toyota Prius prototype with a full perception-localization-planning-control stack and shows superior safety and efficiency in simulations against baselines, as well as real-world tests on a closed track in both autonomous and remote operation modes. This work advances practical VRU navigation by combining topology-based trajectory optimization with remote operation, enabling safer operation in unstructured urban contexts and offering a structured pathway for handling exceptional situations through human oversight.

Abstract

We present a vehicle system capable of navigating safely and efficiently around Vulnerable Road Users (VRUs), such as pedestrians and cyclists. The system comprises key modules for environment perception, localization and mapping, motion planning, and control, integrated into a prototype vehicle. A key innovation is a motion planner based on Topology-driven Model Predictive Control (T-MPC). The guidance layer generates multiple trajectories in parallel, each representing a distinct strategy for obstacle avoidance or non-passing. The underlying trajectory optimization constrains the joint probability of collision with VRUs under generic uncertainties. To address extraordinary situations ("edge cases") that go beyond the autonomous capabilities - such as construction zones or encounters with emergency responders - the system includes an option for remote human operation, supported by visual and haptic guidance. In simulation, our motion planner outperforms three baseline approaches in terms of safety and efficiency. We also demonstrate the full system in prototype vehicle tests on a closed track, both in autonomous and remotely operated modes.
Paper Structure (11 sections, 2 equations, 10 figures, 1 table)

This paper contains 11 sections, 2 equations, 10 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Vehicle System
  4. Prototype Vehicle
  5. Perception
  6. Localization and Mapping
  7. Motion Planning
  8. Control
  9. Remote Operation
  10. Experiments
  11. Conclusions

Figures (10)

  • Figure 1: Safe and efficient vehicle navigation among VRUs.
  • Figure 2: Overview of the demonstrator's architecture. The dashed-colored regions indicate the main components. The remote transmission symbol indicates data being exchanged between the vehicle and the external Remote Control Tower.
  • Figure 3: Visualization of the object detections. The detector detects a predefined set of object classes (e.g. car and pedestrian).
  • Figure 4: LiDAR clustering visualized. It segments any object protruding from the ground plane, including generic objects.
  • Figure 5: The point cloud map used for map-based localization in our demo. A point's hue indicates its height in the z-direction.
  • ...and 5 more figures