Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Geo-locating Road Objects using Inverse Haversine Formula with NVIDIA Driveworks

Mamoona Birkhez Shami, Gabriel Kiss, Trond Arve Haakonsen, Frank Lindseth

TL;DR

The paper tackles the challenge of geolocating road objects from a single monocular camera for ADAS/AV applications. It proposes an end-to-end pipeline that leverages NVIDIA DriveWorks for depth and heading estimation, precise camera calibration, and GNSS-based ground truth via CPOS, culminating in geolocation with the inverse Haversine formula. Across stationary and moving targets in controlled and real-world settings, the method achieves sub-meter accuracy when stationary and under 4 meters at speeds up to 60 km/h within a 15 m radius, demonstrating practical viability. The main contributions include the inverse Haversine-based geolocation algorithm, a detailed DriveWorks calibration protocol, and extensive validation across diverse scenarios, with future work aimed at automation, multi-camera fusion, and digital-twin integration for infrastructure perception.

Abstract

Geolocation is integral to the seamless functioning of autonomous vehicles and advanced traffic monitoring infrastructures. This paper introduces a methodology to geolocate road objects using a monocular camera, leveraging the NVIDIA DriveWorks platform. We use the Centimeter Positioning Service (CPOS) and the inverse Haversine formula to geo-locate road objects accurately. The real-time algorithm processing capability of the NVIDIA DriveWorks platform enables instantaneous object recognition and spatial localization for Advanced Driver Assistance Systems (ADAS) and autonomous driving platforms. We present a measurement pipeline suitable for autonomous driving (AD) platforms and provide detailed guidelines for calibrating cameras using NVIDIA DriveWorks. Experiments were carried out to validate the accuracy of the proposed method for geolocating targets in both controlled and dynamic settings. We show that our approach can locate targets with less than 1m error when the AD platform is stationary and less than 4m error at higher speeds (i.e. up to 60km/h) within a 15m radius.

Geo-locating Road Objects using Inverse Haversine Formula with NVIDIA Driveworks

TL;DR

The paper tackles the challenge of geolocating road objects from a single monocular camera for ADAS/AV applications. It proposes an end-to-end pipeline that leverages NVIDIA DriveWorks for depth and heading estimation, precise camera calibration, and GNSS-based ground truth via CPOS, culminating in geolocation with the inverse Haversine formula. Across stationary and moving targets in controlled and real-world settings, the method achieves sub-meter accuracy when stationary and under 4 meters at speeds up to 60 km/h within a 15 m radius, demonstrating practical viability. The main contributions include the inverse Haversine-based geolocation algorithm, a detailed DriveWorks calibration protocol, and extensive validation across diverse scenarios, with future work aimed at automation, multi-camera fusion, and digital-twin integration for infrastructure perception.

Abstract

Geolocation is integral to the seamless functioning of autonomous vehicles and advanced traffic monitoring infrastructures. This paper introduces a methodology to geolocate road objects using a monocular camera, leveraging the NVIDIA DriveWorks platform. We use the Centimeter Positioning Service (CPOS) and the inverse Haversine formula to geo-locate road objects accurately. The real-time algorithm processing capability of the NVIDIA DriveWorks platform enables instantaneous object recognition and spatial localization for Advanced Driver Assistance Systems (ADAS) and autonomous driving platforms. We present a measurement pipeline suitable for autonomous driving (AD) platforms and provide detailed guidelines for calibrating cameras using NVIDIA DriveWorks. Experiments were carried out to validate the accuracy of the proposed method for geolocating targets in both controlled and dynamic settings. We show that our approach can locate targets with less than 1m error when the AD platform is stationary and less than 4m error at higher speeds (i.e. up to 60km/h) within a 15m radius.
Paper Structure (21 sections, 3 equations, 11 figures, 1 table)

This paper contains 21 sections, 3 equations, 11 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Works
  3. Method
  4. Data
  5. Data Acquisition with Autonomous Vehicle
  6. National Road Database (NVDB)
  7. GNSS Positioning with CPOS Service
  8. Camera Calibration using NVIDIA DriveWorks
  9. Geolocation Estimation
  10. Estimating distance of the target from the camera on the vehicle:
  11. Estimating heading of the target from the car:
  12. Geolocating the target:
  13. Experiments
  14. Results and Discussion
  15. Stationary Vehicle in a Controlled Environment
  16. ...and 6 more sections

Figures (11)

  • Figure 1: Flowchart for the proposed method of geolocation.
  • Figure 2: This figure illustrates the placement of the sensors on the autonomous vehicle at NAPLab, NTNU gusev2022remote.
  • Figure 3: This figure illustrates the placement of the sensor for ground truth data. The left image shows the orange control marker in the NTNU Gloshaugen campus parking lot. The right image shows the measurement for traffic signs.
  • Figure 4: Flowchart for the camera calibration process using NVIDIA DriveWorks
  • Figure 5: The top image shows the placement of the AprilTag targets around the car for extrinsic calibration. The image is taken from NVIDIA DriveWorks documentationNvidiaCalib. The left image in the bottom row shows the validation of the intrinsic calibration of camera 3, and the right image validates the extrinsic calibration of camera 3.
  • ...and 6 more figures