A comparison of neural network, state augmentation, and multiple model-based approaches to online location of inertial sensors on a vehicle is presented that exploits dual-antenna carrier-phase-differential GNSS. The best technique among these is shown to yield a significant improvement on a priori calibration with a short window of data. Estimation of Inertial Measurement Unit (IMU) parameters is a mature field, with state augmentation being a strong favorite for practical implementation, to the potential detriment of other approaches. A simple modification of the standard state augmentation technique for determining IMU location is presented that determines which model of an enumerated set best fits the measurements of this IMU. A neural network is also trained on batches of IMU and GNSS data to identify the lever arm of the IMU. A comparison of these techniques is performed and it is demonstrated on simulated data that state augmentation outperforms these other methods.

Cite and download the paper:
Nick Montalbano, and Todd E. Humphreys "A Comparison of Methods for Online Lever Arm Estimation in GPS/INS Integration," in Proceedings of the IEEE/ION PLANS Meeting, Monterey, CA, 2018.


Recognizing objects in the environment and precisely determining their positions is a fundamental component of autonomous navigation systems. This thesis presents a technique for determining both the locations and the semantic labels of new objects in a scene with respect to a prior three- dimensional (3D) map of the scene. This work aims to reduce object recognition errors in cluttered environments by isolating new objects from the known background by correlating features de- tected in a new photo with feature points that constitute the 3D map. Such isolation enables a neural network trained to recognize an enumerated set of objects to focus narrowly upon those portions of images that contain new objects instead of having to process the whole scene. As a result, changes in a prior map can be rapidly detected and semantically labeled, allowing for confident navigation within the ever-evolving cluttered environment. Using multiple images ob- tained from varying camera poses, the globally-referenced 3D positions of changes in the scene can be determined with multiple-view geometry techniques.

Cite and download the paper:
Siddarth Kaki, Todd E. Humphreys, Maruthi Akella "Exploiting a Prior 3D Map for Object Recognition

A new method is developed for integer ambiguity resolution in carrier-phase differential global navigation satellite system (CDGNSS) positioning. The method is novel in that it is simultaneously (1) data-driven, (2) generalized to include partial ambiguity resolution, and (3) amenable to a full analytical characterization of the prior probabilities of correctly- and incorrectly-resolved ambiguities. The technique is termed generalized integer aperture bootstrapping, or GIAB. A full development of GIAB is provided herein, including sizing its integer aperture to usually produce a higher prior probability of full ambiguity resolution than comparable existing methods. In Monte-Carlo simulations, GIAB is shown to provide nearly optimal ambiguity resolution success rates of full ambiguity resolution for relevant integrity requirements under strong models while enabling partial ambiguity resolution.

Cite and download the paper:
G. Nathan Green, and Todd E. Humphreys "Data-Driven Generalized Integer Aperture Bootstrapping for High-Integrity Positioning," To appear.


A full analysis of position domain integrity is carried out for the recently-introduced Generalized Integer Aperture Bootstrapping (GIAB) technique, a data-driven method for resolving and validating GNSS carrier-phase integer ambiguities suitable for high-integrity, safety-critical systems. The analysis can be extended to all integer aperture (IA) techniques that are generalized in the sense of allowing partial integer fixing. It is shown that generalized IA methods produce relative position (baseline) estimates that suffer from non-negligible biases. Key conditional distributions of the baseline computed from GIAB- validated ambiguities are rigorously derived for both full and partial ambiguity resolution. These distributions enable evaluation of the a posteriori risk from bias in the GIAB baseline estimate. Compared to EPIC, the state-of-the-art high-integrity algorithm, GIAB is shown to satisfy tighter integrity requirements for the same measurement model.

Cite and download the paper:
G. Nathan Green, and Todd E. Humphreys "Position Domain Integrity Analysis for Generalized Integer Aperture Bootstrapping," To appear.


A system developed at The University of Texas for low-cost precise urban vehicular positioning is demonstrated to achieve a probability of correct integer fixing greater than 96.5% for a probability of incorrect integer fixing surely less than 2.3% and likely less than 1%. This is demonstrated using data captured during 3.4 hours of driving on a repeating urban test route over three separate days. The results are achieved without any aiding by inertial or electro-optical sensors. Development and evaluation of the unaided GNSS-based precise positioning system is a key milestone toward the overall goal of combining precise GNSS, vision, radar, and inertial sensing for all-weather high-integrity precise positioning for automated and connected vehicles. The system described and evaluated herein is composed of a densely-spaced reference network, a software-defined GNSS receiver whose processing can be executed on general-purpose commodity hardware, and a real-time kinematic (RTK) positioning engine. All components have been tailored in their design to yield competent sub-decimeter positioning in the mobile urban environment. A performance sensitivity analysis reveals that navigation data bit prediction on the GPS L1 C/A signals is key to high-performance urban RTK positioning.

Cite and download the paper:
Todd E. Humphreys, Matthew Murrian, and Lakshay Narula "Low-cost Precise Vehicular Positioning in Urban Environments," in Proceedings of the IEEE/ION PLANS Meeting, Monterey, CA, 2018.


More Articles...