Image Data-Driven-Deep Learning in Geosystems: Exploratory Investigation into the Stability of Retaining Walls


PI:  Zhen Liu


Project Period:  09.01.17 – 08.31.19

The project is proposed to introduce the most recent breakthroughs in computer vision attributed to deep learning to address a rarely discussed yet urgent issue in most engineering disciplines: how to analyze the explosively increasing image data including images and videos, which can hardly be analyzed with traditional methods? Therefore, the proposed work is very time-sensitive. The core concepts enabling the breakthroughs in image recognition and adopted in AlphaGo, i.e., deep convolutional neural nets, will be used to explore the possibility to accurately assessing the safety of retaining walls with image data. The analysis will be validated against traditional methods including limit analysis and numerical simulation. The research is hypothesis-driven and rationally built on our preliminary study, which clearly supports the hypothesis and shows the high potential of the proposed study. In addition to further proving the power of the proposed concept, this collaborative effort between geotechnical engineering and computer science will understand the data and deep learning in geotech analysis, relate image patterns to physical mechanisms, and investigate two key potential issues associated with the methodology.

The proposed work will be organized into three tasks. Task 1 is to understand how the geotechnical data and analysis results can be related to the input and output of deep learning using convolutional nets. This is a key to the successful application of deep learning in geotech. In Task 2, eight factors in three categories, i.e., geology (soil property), topography (geometry) and boundary conditions, will be evaluated regarding their influence on the safety factors. The relationships between these factors, their roles and significance in deep learning, and correlation between the patterns identified by the CNN and the physical mechanisms will be compared and revealed to connect the deep learning to the existing geotech knowledge pool. In Task 3, two additional key aspects representing the unknowns and possible weaknesses of the machine learning method will be analyzed: model robustness and extrapolation. Testing data prepared in ways different from the input data, i.e., appearance and ranges, will be used to assess and improve these two aspects.

Zhen Liu
Zhen Liu
Shiyan Hu
Shiyan Hu


DRU Collaborative Research: Understanding Mental Models of Expertise in Construction Management using Interactive Adaptive Simulations

Sponsor: National Science Foundation

PI: Amlan Mukherjee

This research uses recent advances in simulations and data analysis techniques to investigate the cognitive and engineering aspects of decision making in complex dynamic construction management scenarios. Expertise plays a crucial role in managing crisis scenarios that call for critical decision making under constraints of time, resource, and rapidly unfolding events. An example of such a crisis scenario is managing complex heavy construction projects. In such scenarios, effective decision making requires knowledge of complex inter-relationships between several simultaneous events and preparing for the uncertainty and risks arising from feedbacks in time and space. Such knowledge is inductively constructed by assimilating and organizing experiential knowledge into patterns of information that are difficult to formalize or analytically perceive. The researchers propose to investigate the dynamics and variation of such cognitive knowledge organization patterns, or mental models, of decision making, specifically among construction managers.

The goal of this research effort will be to use an interdisciplinary approach to understand how expert and novice construction managers differ in their knowledge organization, information processing, risk assessment, and decision making in construction management crisis scenarios. Interactive, adaptive simulations of critical construction scenarios will be developed in collaboration with construction management firms, and expert and novice construction managers will be tested in them to capture human-subject interaction data that will be analyzed to develop mental models of expertise. In addition, an instructional interface will be integrated into the simulation using pedagogical agents, and it will be deployed in the construction management curriculum to test its effectiveness as a training environment for novice decision makers. This will also allow the researchers to investigate how novices construct knowledge in simulated training environments.

Amlan Mukherjee
Amlan Mukherjee

2011 Michigan Seat Belt Observation Surveys

Sponsor: Michigan State Police

PI: John Hill

This report summarizes the findings of the statewide seatbelt survey completed in June following the May 2011 Click it or Ticket enforcement campaign.  191 Intersections were surveyed.  The overall seat belt use rate was 94.5%.  For the second survey in August, the seat belt use rate fell to 93.5%.  Additionally, information regarding cell phone use is contained in this report.  Cell phone use rate among drivers was 8.6% in June and 10.3% in August.  All data is organized by vehicle type, age, gender, race, and day and time.  Additionally, logistic regression models have been provided to identify the statistical significance of differences in seat belt use and cell phone use based on these characteristics.  Wayne County showed a significant increase in seat belt usage, and may be a case study for best practices.  With regard to cell phone usage, the survey identified Hispanics and females age 16‐29 as the primary users of cell phones while driving.


Beyond Traffic Innovation Center (BTIC)


PI:  Pasi Lautala

The close partnerships between diverse entities on MTU’s campus allow Michigan Tech to serve the transportation field across many areas. Innovations and research conducted under entities such as the Michigan Tech Research Institute, Advanced Power Systems Research Center, and numerous laboratories can be integrated into our academic programs by departments, but also disseminated to practitioners and public stakeholders through our Center for Technology and Training (CTT), Tribal Technical Assistance Program (TTAP) and Center for Science and Environmental Outreach (CSEO).

Michigan Tech has a strong and versatile academic program in transportation. Our Department of Civil and Environmental Engineering offers BS, MS, and PhD concentrations in transportation. In addition, we house one of the few Rail Transportation Programs in the nation and perhaps the only Minor in Rail Transportation available today. There also are certificate programs in Hybrid Electric Drive Vehicle Engineering, Advanced Electric Power Engineering and a graduate certificate in Automotive Systems, plus numerous opportunities for undergraduate and graduate level transportation research in various disciplines. These include our innovative Enterprise Programs for undergraduate research in areas, such as alternative energy, hybrid electric vehicles and wireless communications.

In addition to our academic programs, we offer professional training and outreach in transportation topics through a variety of centers. CTT is a transportation training and outreach center focusing on practitioner training, technical assistance, and best practices that enhance business and technical practices for state and local agencies. Their training typically reaches 5,000 participants annually with over 24,000 contact hours. TTAP offers similar services to American Indian governments and communities in the 31 states bordering and east of the Mississippi River (Federal Lands Eastern Region), reaching 1,100 annual participants and 4,200 contact hours. Finally, the Michigan Tech Mobile Lab is a fully outfitted mobile laboratory that partners with government, industry, and nonprofit organizations to deliver HEV (hybrid-electric vehicle) education, outreach, and research across the nation.

Michigan Tech has been instrumental in organizing regional and national conferences and workshops, such as the annual Michigan Rail Conference founded by Michigan Tech in collaboration with the Michigan Department of Transportation (MDOT) and the National Tribal Transportation Conference. Michigan Tech also collaborated in a workshop funded by FHWA with participants from state DOTs like Caltrans, city government, European representatives, and industry to “Address Infrastructure Life Cycle Inventory Data Needs: Supporting Sustainable Decision-Making for Civil Infrastructure Using EPDs.” Dr. John Harvey from the University of California Pavement Research Consortium (UCPRC) was a close collaborator on the effort. In 2016, we also initiated the Exploring Next Generation IN-vehicle INterfaces Consortium (ENGIN) and related speaker series.

Michigan Tech educates and encourages K-12 students to advance in the transportation field through several youth events and summer programs. These include the National Summer Transportation Institute (NSTI), Rail and Intermodal Summer Youth Program (SYP), Women in Automotive Engineering, and Human Factors Engineering programs. We also host the annual Clean Snowmobile Challenge at Michigan Tech’s Keweenaw Research Center.

Michigan Tech is engaged in national and regional decision making through participation in and leadership of committees. Some examples include the Chairmanship of the TRB AR040 Freight Rail Committee and involvement in AASHTO Subcommittee on Materials and FHWA Sustainable Pavements Technical Working Group. Michigan Tech is also represented in the seven-member State of Michigan Commission for Supply Chain and Logistics Collaboration, and our faculty/staff has obtained national and regional awards, such as the Wootan Award received by Timothy Colling and the 2015 WisDOT Tribal Excellence Award from the Wisconsin Department of Transportation received by John Velat.

While not located in one of the 11 megaregions identified in the Beyond Traffic 2045, our location in rural Michigan makes us ideal in addressing trends and challenges faced by rural transportation, and we have worked closely with Michigan Department of Transportation and local governments in these issues. However, Michigan Tech’s leadership is not restricted to rural aspects. Many of our activities have broader impacts, such as the RoadSoft asset management software developed at Michigan Tech and used in the Michigan and several others states by rural and urban counties alike.

Michigan Tech is directly involved in addressing several trends/challenges identified in the Beyond Traffic document. We have conducted numerous studies related to the challenges faced in freight transportation in our region, conducted Life-Cycle Analysis (LCA) to evaluate environmental impacts of different transportation materials and related solutions, are actively involved in emissions research for various engines, are one of the leaders in advancing open source 3D printing, have on-going projects related to V2V and automated vehicle research, and have had various projects related to alternative energies for transportation, especially in the biomass supply chains for biofuel development. Michigan Tech researchers and educators accomplish this innovative work through collaborative thinking and working across departments and disciplines, which allows us to tackle large-scale projects that require a diverse skill set.


Pasi Lautala
Pasi Lautala

Identifying Cost and Funding Alternatives for Equipping Operating While Intoxicated Offenders with Ignition Interlock Devices

Sponsor:  Wisconsin Department of Transportation

PI:  John Hill

Ignition Interlock Devices (IID) have been used in multiple states to deter repeat operating while intoxicated (OWI) offenders from further offenses.  It has been found in the state of California that a group of offenders who had an order to use an IID had a reduction in future crash rates of 24 percent, while slight changes were seen in those that were not ordered the use of an IID.  Those drivers who installed the IID also had a lower rate of future DUI convictions.   A major issue with IIDs has been compliance.  Studies have shown that as few as 10% of drivers convicted of OWI and ordered to install an IID device, actually do so.

Research Objectives
– Identify and characterize OWI offenders in Wisconsin  and develop a 10 year forecast of overall
OWI arrests and arrests of particular high risk offenders.
– Estimate IID implementation costs based on the forecast model developed and determine the
overall affordability of IID devices for OWI offenders
– Identify potential funding sources to increase IID installations.

Information regarding over 200,000 OWI offenders in Wisconsin from 2005-2009 was analyzed.  Using a logistic regression model, characteristics of OWI offenders who were likely to repeat the offense within 1 or 2 years were identified.  Additionally, a 10 year forecast model was developed which was based on annual historical arrest levels over the past 20 years, as well as monthly vehicle miles travelled an d economic data over the past 5 years.  Using this information, along with cost data collected from interviewing IID manufacturers, the overall cost to equip the vehicles of future OWI offenders was collected.  Income levels of IID offenders were also modeled using an exponential distribution to determine what proportion of offenders may be unable to afford an IID.  Finally, an analysis of potential funding sources was conducted to determine if supplementing the cost of IIDs might be a feasible means of increasing IID order compliance.