Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

VIPS-Odom: Visual-Inertial Odometry Tightly-coupled with Parking Slots for Autonomous Parking

Xuefeng Jiang, Fangyuan Wang, Rongzhang Zheng, Han Liu, Yixiong Huo, Jinzhang Peng, Lu Tian, Emad Barsoum

TL;DR

VIPS-Odom tackles precise underground parking localization by tightly fusing visual-inertial measurements with semantic parking-slot observations detected in BEV. It combines frontend enhancements (parking-slot corner features merged with Shi-Tomasi points) and backend semantic constraints (PS registration terms within a sliding-window optimization) to reduce drift. The system uses YOLOv5 for PS detection, SORT for robust PS state tracking, BEV via IPM, and a multi-sensor backend that includes WSS along with IMU data, achieving improved accuracy over baselines in both short- and long-distance parking scenarios. The experimental results on real vehicles show the method's effectiveness and stability, with potential future extensions including loop closure and additional semantic objects.

Abstract

Precise localization is of great importance for autonomous parking task since it provides service for the downstream planning and control modules, which significantly affects the system performance. For parking scenarios, dynamic lighting, sparse textures, and the instability of global positioning system (GPS) signals pose challenges for most traditional localization methods. To address these difficulties, we propose VIPS-Odom, a novel semantic visual-inertial odometry framework for underground autonomous parking, which adopts tightly-coupled optimization to fuse measurements from multi-modal sensors and solves odometry. Our VIPS-Odom integrates parking slots detected from the synthesized bird-eye-view (BEV) image with traditional feature points in the frontend, and conducts tightly-coupled optimization with joint constraints introduced by measurements from the inertial measurement unit, wheel speed sensor and parking slots in the backend. We develop a multi-object tracking framework to robustly track parking slots' states. To prove the superiority of our method, we equip an electronic vehicle with related sensors and build an experimental platform based on ROS2 system. Extensive experiments demonstrate the efficacy and advantages of our method compared with other baselines for parking scenarios.

VIPS-Odom: Visual-Inertial Odometry Tightly-coupled with Parking Slots for Autonomous Parking

TL;DR

VIPS-Odom tackles precise underground parking localization by tightly fusing visual-inertial measurements with semantic parking-slot observations detected in BEV. It combines frontend enhancements (parking-slot corner features merged with Shi-Tomasi points) and backend semantic constraints (PS registration terms within a sliding-window optimization) to reduce drift. The system uses YOLOv5 for PS detection, SORT for robust PS state tracking, BEV via IPM, and a multi-sensor backend that includes WSS along with IMU data, achieving improved accuracy over baselines in both short- and long-distance parking scenarios. The experimental results on real vehicles show the method's effectiveness and stability, with potential future extensions including loop closure and additional semantic objects.

Abstract

Precise localization is of great importance for autonomous parking task since it provides service for the downstream planning and control modules, which significantly affects the system performance. For parking scenarios, dynamic lighting, sparse textures, and the instability of global positioning system (GPS) signals pose challenges for most traditional localization methods. To address these difficulties, we propose VIPS-Odom, a novel semantic visual-inertial odometry framework for underground autonomous parking, which adopts tightly-coupled optimization to fuse measurements from multi-modal sensors and solves odometry. Our VIPS-Odom integrates parking slots detected from the synthesized bird-eye-view (BEV) image with traditional feature points in the frontend, and conducts tightly-coupled optimization with joint constraints introduced by measurements from the inertial measurement unit, wheel speed sensor and parking slots in the backend. We develop a multi-object tracking framework to robustly track parking slots' states. To prove the superiority of our method, we equip an electronic vehicle with related sensors and build an experimental platform based on ROS2 system. Extensive experiments demonstrate the efficacy and advantages of our method compared with other baselines for parking scenarios.
Paper Structure (20 sections, 10 equations, 7 figures, 8 tables)

This paper contains 20 sections, 10 equations, 7 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Multi-sensor SLAM
  4. Semantic SLAM for Autonomous Parking
  5. Methodology
  6. Sensor Equipment & Calibration
  7. BEV Image Generation
  8. Optimization Formulation
  9. Parking Slot (PS) Detection & Management
  10. Frontend Integration with PS
  11. Backend Optimization with PS
  12. Experiments
  13. Experiment Setup
  14. Experimental Results
  15. Ablation Study
  16. ...and 5 more sections

Figures (7)

  • Figure 1: Difficulties of traditional methods in typical parking environments.
  • Figure 2: A brief illustration of our sensor configuration.
  • Figure 3: Framework of VIPS-Odom. Our system integrates parking slot observation with measurements from IMU/WSS together for tightly-coupled optimization. By introducing constraint from parking slot observation, the system achieves higher accuracy and stability.
  • Figure 4: Feature points acquisition of frontend. (a) is parking slot detection results on the BEV image synthesized via IPM. (b) is parking slot feature points in the front-view fisheye, which is obtained by inverse projection.
  • Figure 5: An illustration of our backend optimization. Note that in the BEV frame, the center point of parking slot #2 is closer to the BEV center point than other slots (i.e. #1,#3, and #4).
  • ...and 2 more figures