Off-campus Michigan Tech users: To download campus access theses or dissertations, please use the following button to log in with your Michigan Tech ID and password: log in to proxy server

Non-Michigan Tech users: Please talk to your librarian about requesting this thesis or dissertation through interlibrary loan.

Date of Award


Document Type

Campus Access Master's Thesis

Degree Name

Master of Science in Electrical and Computer Engineering (MS)

Administrative Home Department

Department of Electrical and Computer Engineering

Advisor 1

Jeremy P. Bos

Committee Member 1

Anthony Pinar

Committee Member 2

Thomas Oommen


Measuring the depth of ruts caused by vehicles on different terrains is very important as it provides information that can be used for terrain assessment, timely maintenance, vehicle traversability, ecological impact, and others. Traditionally, rut depth is measured manually but recently algorithms have been developed to automate the measurement process using different sensors like cameras, laser scanners, etc. However, more work needs to be done on comparing sensors based on various factors like cost and accuracy.

In this thesis, I compare three sensors, an Intel RealSense Depth Camera D435i, a Hokuyo UST-10LX laser scanner, and a Velodyne VLP-16 Light Detection and Ranging (LiDAR), for rut depth measurement and develop an algorithm for autonomous rut depth measurement. The algorithm to measure rut depth is developed while making sure that it mimics the standard rut depth measurement technique where a plank is placed across the rut and the perpendicular distance from the bottom of the rut to the plank is measured manually. The sensors are compared on performance metrics like accuracy, cost, and computation time. Ruts were developed by driving Clearpath's Husky A200 UGV on snow and rut depth was measured manually using the standard technique and using the above-mentioned sensors mounted on the robot. The development and testing of the algorithm were done on the Intel NUC computer with Intel i7 6770 processor with 32 GB RAM.

It was observed from testing results that Veldoyne VLP-16 3D LiDAR had the least standard deviation followed by the Hokuyo UST-10LX 2D LiDAR, and the Realsense D435i had the largest standard deviation. The VLP-16 had the smallest mean error followed by the Intel Realsense D435i depth camera and the Hokuyo had the largest mean error. The VLP-16 costs \$4000, the Hokuyo costs \$1200, and the depth camera costs \$400 currently, the cost difference of the sensors presents a trade-off that depends on the requirements of the application. The future scope of this work will be to measure rut depth in real-time so that the developed algorithm can be combined with an Automated Dynamic Cone Penetrometer (ADCP) for efficient soil testing of an unknown area. The developed algorithm has been tested in ruts created by driving in snow, and the performance of the sensors might change when measuring soil rut depth because soil ruts have higher edges due to soil accumulation at the edges. Also, different rut conditions can also affect sensor performance, for example, reflectivity from the water might affect the camera performance when, measuring rut depth in water-filled ruts, and so another future work will be to conduct testing on different terrains and soil types.