ZJZachary D Jeffries

Zachary D Jeffries

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Zachary D Jeffries has a total of 4 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    EVALUATION OF LIDAR UNCERTAINTY AND APPLICATIONS TOWARDS SLAM IN OFF-ROAD ENVIRONMENTS

    by Zachary D Jeffries

    Safe and robust operation of autonomous ground vehicles in all types of conditions and environment necessitates complex perception systems and unique, innovative solutions. This work addresses automotive lidar and maximizing the performance of a simultaneous localization and mapping stack. An exploratory experiment and an open benchmarking experiment are both presented. Additionally, a popular SLAM application is extended to use the type of information gained from lidar characterization, demonstrating the performance gains and necessity to tightly couple perception software and sensor hardware. The first exploratory experiment collects data from child-sized, low-reflectance targets over a range from 15 m to 35 m. The targets are placed in close proximity with respect to the azimuthal sweep of scanning lidars in an effort to stress the sensor’s gain controllers. Results are presented for range variance with respect to target distance, the propagation of beam spot size, an analysis on the resolvability of target width, and a convolutional analysis of the interaction between the incident beam and the leading and trailing edges of the targets. The range distributions are not normal, and often appears as an ‘H’ when viewed from the top-down. The second experiment was an open benchmarking experiment. Data was collected using calibrated child-size targets. The targets have lambertian reflections and are low-reflectance, and calibrated to maintain those characteristics across the range of lidar wavelengths tested. The nearest target during this experiment was at 5 m while the furthest was at 200 m. A benchmarking test with the targets alone revealed range biases varying from −4.21 cm to 8.16 cm. Average range precision was between 1.33 cm and 7.57 cm. A second benchmark was taken with confusers placed adjacent to the targets. These confusers were retroreflective and varied in size. Presence of the confusers increased the range of average range bias between −4.96 cm and 8.42 cm. The precision was reduced, having a range of 2.22 cm to 7.57 cm. Furthermore, the number of detections or points on target, were reduced, sometimes by more than 20%. The third work presented is LIOSAM-LH. It is an extension, using lidar heuristics, of the state-of-the-art lidar-inertial SLAM framework, LIO-SAM. LIOSAM-LH uses GTSAM to provide a factor-graph based odometry solution. Two key contributions, a correction layer and a factor-augmentation layer, enable increased map accuracy, more stable odometry tracks, and reduced computation times. Specifically, the corrections layer reduces artificial inflation of objects in the pointcloud map by greater

  • Open Access

    Toward open benchmark tests for automotive lidars, year 1: static range error, accuracy, and precision

    by Zach Jeffries, Jeremy P. Bos, Paul McManamon, Charles Kershner, Akhil Kurup

    This paper describes the initial results from the first of 3 years of planned testing aimed at developing methods, metrics, and targets necessary to develop standardized tests for these instruments. Here, we evaluate range error accuracy and precision for eight automotive grade lidars; a survey grade lidar is used as a reference. These lidars are tasked with detecting a static, child-sized, target at ranges between 5 and 200 m. Our target, calibrated to 10% reflectivity and Lambertian, is a unique feature of this test. We find that lidar range precision is in line with the values reported by each manufacturer. However, we find that maximum range and target detection can be negatively affected by presence of an adjacent strong reflector. Finally, we observe that design trade-offs made by each manufacturer lead to important performance differences that can be quantified by tests such as the ones proposed here. This paper also includes some lessons learned, planned improvements, and discussion of future iterations of this activity.