Vehicle lane keeping through pavement-assisted passive sensing

Dahal, Sachindra

Permalink

https://hdl.handle.net/2142/115711 Copy

Description

Title

Vehicle lane keeping through pavement-assisted passive sensing

Author(s)

Dahal, Sachindra

Issue Date

2022-04-19

Director of Research (if dissertation) or Advisor (if thesis)

Roesler, Jeffery R

Doctoral Committee Chair(s)

Roesler, Jeffery R

Committee Member(s)

• Al-Qadi, Imad L
• Popovics, John S
• Mitra, Sayan

Department of Study

Civil & Environmental Eng

Discipline

Civil Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

Lane keeping, Autonomous vehicles, ADAS, concrete pavement, Vehicle to Infrastructure

Abstract

Autonomous vehicles (AV) and advanced driver-assistance systems (ADAS) rely on sensors to offer multiple safety benefits for drivers and road agencies. As a result, the number of cars using sensors to enhance driving safety has increased rapidly and continues to rise. Determining the vehicle lateral position within the lane (lane keeping) is one of the safety-critical applications. However, maintaining the lateral position of a vehicle within a lane over a wide range of conditions is a challenge, particularly in adverse weather conditions when lane markings are occluded. For large scale adaptation of ADAS/AVs without compromising safety, vehicle-to-infrastructure sensing capabilities are necessary for redundancy, especially during severe weather conditions. A method to create a continuous electromagnetic (EM) signature on the roadway was proposed using EM materials compatible with existing paving materials and construction methods to assist in maintaining the lateral position of an AV or vehicle with ADAS. The passive EM signature is detected by sensors on the vehicle to determine their lateral position within the lane irrespective of the visibility of lane markings. Laboratory testing of the proposed concept was performed on concrete slab specimens that allowed the insertion of concrete prisms containing various EM materials. An induction-based eddy-current sensor and magnetometers were deployed to detect the passive EM signatures. The induction-based sensing method was found to be inferior to the magnetometer array. Ultimately, the detected EM signals with magnetometers were compared to evaluate the effects of sensor height above the concrete surface, type of EM materials, EM-material volume, material shape, and volume of EM concrete prisms. A layer of up to 5 cm (2 inch) of water, ice, snow, or sand was placed between the sensor and the notched concrete slab with the different EM concrete prisms to represent adverse weather conditions. Experimental results showed that sensor height, EM-material volume, EM dosage, type of the EM material, and shape of the EM material in the concrete prism were significant attenuators of the EM signal and must be engineered properly. A minimum peak of EM signature of 147nT was detected in the lab using magnetometer sensor array. Presence of adverse surface conditions had a negligible effect relative to standard conditions indicating the robustness of the proposed method. Asphalt and concrete materials have similar EM signal which can be significantly enhanced by embedding EM materials in new pavements or rehabilitating existing roadway surfaces. To determine the vehicle's position within the lane using the proposed method, field testing was performed by placing concrete prisms containing EM materials at the centerline of a 150 ft. long and 12 ft. wide lane. A vehicle was equipped with a custom magnetometer sensor array, front viewing camera, and cameras on both side mirrors looking down at the lane marking. The magnetometer sensors in gradiometer configuration were mounted on the vehicle's front bumper and detected the EM signature from concrete prisms containing different types and depths of EM material from the lab testing. The ground truth was established as the average vehicle location detected by two side mirror-mounted cameras that looked directly on the lane markings. The vehicle's lateral position with time was then estimated from the magnetometer sensor array and camera and compared with the ground truth to calculate the error. For any time step, an error of either sensor was considered the absolute difference between the ground truth and the position calculated by the respective sensor. Statistical analysis of error for a total of 42 vehicle passes corresponding to driving approximately 6,300 ft, was done in various weather conditions and motion types within the lane (straight and meandering). The results indicated that in the normal condition, the camera and magnetometer had a comparable error of 3.3 and 4 cm, respectively. The error of the camera increased to 10.4 cm, while the magnetometer had an error of 3.1 cm when less than 2.5 cm of snow partially occluded the lane marking. With more than 5 cm (2 inch) of snow on the lane, the camera could not perceive lane markings, failing to determine lateral position of the vehicle. The proposed magnetometer sensor detected the vehicle's lateral position with an average error of 4.6 cm, which was statistically similar to normal conditions. The pavement-based EM signature and lateral position results provided a method to complement existing sensors’ limitations in AVs and ADAS for effective lane keeping during normal and adverse weather conditions. Finally, a method was developed to create and erase EM signature by introducing certain chemical in powder form to the near surface of a concrete sample. Through a post-construction treatment of concrete, a detectable EM signature can be created that is easily detectable and subsequently erased if necessary. The proposed framework, materials, and sensing method opens a new paradigm in pavement design where roadways in the future will not only be designed for physical properties but also for strategic-placed EM properties for enhanced pavement-vehicle interaction.

Graduation Semester

2022-05

Type of Resource

Thesis

Copyright and License Information

Copyright 2022 Sachindra Dahal

Owning Collections