Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Generalized Control Revision Method for Autonomous Driving Safety

Zehang Zhu, Yuning Wang, Tianqi Ke, Zeyu Han, Shaobing Xu, Qing Xu, John M. Dolan, Jianqiang Wang

TL;DR

The paper addresses autonomous driving safety by adding a generalized control revision module based on Dynamic Control Barrier Functions (D-CBF) that can handle heterogeneous perception outputs. It introduces a perception data conversion layer to unify vectorized bounding boxes and occupancy grid maps and a unified traffic-element constraint representation that models both dynamic obstacles and road topology. The control revision is solved as a quadratic program that minimally adjusts upstream planner output while enforcing barrier constraints and road boundaries, enabling safe operation across diverse planning backbones and road topologies. Validations across CARLA, SUMO, OnSite, and a real-world MCCT platform demonstrate reduced accidents and safer maneuvers in complex traffic scenarios, highlighting practical feasibility and scalability of the approach.

Abstract

Safety is one of the most crucial challenges of autonomous driving vehicles, and one solution to guarantee safety is to employ an additional control revision module after the planning backbone. Control Barrier Function (CBF) has been widely used because of its strong mathematical foundation on safety. However, the incompatibility with heterogeneous perception data and incomplete consideration of traffic scene elements make existing systems hard to be applied in dynamic and complex real-world scenarios. In this study, we introduce a generalized control revision method for autonomous driving safety, which adopts both vectorized perception and occupancy grid map as inputs and comprehensively models multiple types of traffic scene constraints based on a new proposed barrier function. Traffic elements are integrated into one unified framework, decoupled from specific scenario settings or rules. Experiments on CARLA, SUMO, and OnSite simulator prove that the proposed algorithm could realize safe control revision under complicated scenes, adapting to various planning backbones, road topologies, and risk types. Physical platform validation also verifies the real-world application feasibility.

A Generalized Control Revision Method for Autonomous Driving Safety

TL;DR

The paper addresses autonomous driving safety by adding a generalized control revision module based on Dynamic Control Barrier Functions (D-CBF) that can handle heterogeneous perception outputs. It introduces a perception data conversion layer to unify vectorized bounding boxes and occupancy grid maps and a unified traffic-element constraint representation that models both dynamic obstacles and road topology. The control revision is solved as a quadratic program that minimally adjusts upstream planner output while enforcing barrier constraints and road boundaries, enabling safe operation across diverse planning backbones and road topologies. Validations across CARLA, SUMO, OnSite, and a real-world MCCT platform demonstrate reduced accidents and safer maneuvers in complex traffic scenarios, highlighting practical feasibility and scalability of the approach.

Abstract

Safety is one of the most crucial challenges of autonomous driving vehicles, and one solution to guarantee safety is to employ an additional control revision module after the planning backbone. Control Barrier Function (CBF) has been widely used because of its strong mathematical foundation on safety. However, the incompatibility with heterogeneous perception data and incomplete consideration of traffic scene elements make existing systems hard to be applied in dynamic and complex real-world scenarios. In this study, we introduce a generalized control revision method for autonomous driving safety, which adopts both vectorized perception and occupancy grid map as inputs and comprehensively models multiple types of traffic scene constraints based on a new proposed barrier function. Traffic elements are integrated into one unified framework, decoupled from specific scenario settings or rules. Experiments on CARLA, SUMO, and OnSite simulator prove that the proposed algorithm could realize safe control revision under complicated scenes, adapting to various planning backbones, road topologies, and risk types. Physical platform validation also verifies the real-world application feasibility.
Paper Structure (15 sections, 14 equations, 11 figures, 1 table)

This paper contains 15 sections, 14 equations, 11 figures, 1 table.

Table of Contents

  1. INTRODUCTION
  2. Preliminaries and background
  3. Methods
  4. Perception Data Conversion
  5. Traffic Element Constraint Representation
  6. Integrating CBF
  7. Experiments and Validation
  8. Experiment Setting
  9. Results
  10. Adaption on various planning backbones
  11. Adaption on various road topologies and risk types
  12. Test on continuous traffic flow scenes
  13. Ablation study of perception data conversion
  14. Real-world Platform Validation
  15. CONCLUSIONS

Figures (11)

  • Figure 1: A diagram of the control revision module.
  • Figure 2: The framework of the proposed control revision method.
  • Figure 3: Failure of bounding box to represent irregular shaped obstacles.
  • Figure 4: A diagram of supplementary perception conversion procedures.
  • Figure 5: Adaption of the control revision module on various planning backbones and simulators.
  • ...and 6 more figures