Walbridge construction company makes jobsite safety a top priority on all of their projects, including their two active jobs on U-M’s North Campus. Students on the Walbridge team will use 360° images from jobsites to develop a computer vision model that detects safety non-compliance on worksites.
Abstract:
Walbridge is an ENR Top 50 Construction Company that desires to lead their industry in maintaining the safest possible worksites. Our #1 Core Value is Safety: “Think, demand and deliver safety in all aspects of our business.”
According to the US Bureau of Labor Statistics, in 2021, nearly 1 in 5 workplace deaths occurred in the construction industry, with just over one third of those being due to falls, slips, and trips. Of these, almost all were from falls to a lower level. Safety policies and procedures aim to avoid death and related injuries, but policies are only as strong as compliance on the worksite.
Students on this team will develop a computer vision system that detects safety non-compliances on construction worksites. They will utilize a 360° video feed capture to scan work areas, and train a model to identify safety hazards and appropriate safety measures. Their accurate model will be delivered within an end-to-end process to collect a feed of construction sites, download video, process and identify potential safety hazards, and report the results to the users. The first applications will concentrate on detecting fall hazards and correctly designed safety railings around holes or raised walkways.
Impact:
This improved automated system could allow for more frequent, or even nearly-continuous, evaluation of work sites, resulting in a reduction of injuries and fatalities among construction workers.
Scope:
Minimum Viable Product Deliverable (Minimum level of success)
- Document the capabilities of current technology solutions and evaluate user needs through an extensive review of construction industry best practices, safety regulations, academic literature (e.g., vision recognition, video processing, data augmentation techniques, relevant application cases), patents, review of any similar commercial solutions, etc.
- Working with the mentors, identify and prioritize use cases, in addition to fall identification, based on impact and likeliness for success.
- Determine the most viable modeling techniques given the dataset available and accuracy requirements. Develop the ground truth dataset combining material from various sources, including student labeled examples and augmented data.
- Develop infrastructure for data acquisition: video collection, data storage, post processing, data augmentation, and, finally, model development.
- Demonstrate a working prototype (data acquisition, model, decision) for one use case, collect feedback and determine points for improvement.
Expected Final Deliverable (Expected level of success)
- Deliver a refined prototype model (including operational instructions) that can be validated in field environment.
- Validate the prototype by testing in a live construction environment.
- Provide a strategic roadmap, incorporating points for refinement, and a setting out a strategy to produce a minimum viable product for the construction industry.
Stretch Goal Opportunities: (High level of success)
- Implement an extended field test.
- Evaluate worker acceptance of the technology.
- Add a second use case for common safety violations.
- Demonstrate the effectiveness of the model in challenging environments (lower light, rain, etc.).
Below are the skills needed for this project. Students with the following relevant skills and interest, regardless of major, are encouraged to apply! This is a team based multidisciplinary project. Students on the team are not expected to have experience in all areas, but should be willing to learn and will be asked to perform a breadth of tasks throughout the two semester project.
General Programming (3 Students)
Specific Skills: Creation of integrated Computer Vision tool (image acquisition, database structure, model implementation, and basic UI)
EECS 281 (or equivalent) is required
All team members are expected to develop Computer Vision knowledge and skills.
Likely Majors: CS, DATA
Computer Vision (3 Students)
Specific Skills: Implementation and tuning of computer vision algorithms, video processing, data augmentation techniques.
EECS 281 (or equivalent) is required
Students should be either co-enrolled or have completed coursework/applied project work in machine/computer vision.
Likely Majors: CS, ROB, DATA
Occupational Safety / Human Factors (1 Students)
Specific Skills: Industrial health and safety processes, human factors design, design for safety.
Students must be able to participate in data preparation/evaluation, and simple technical development of the tool. Ideally, students will have completed a first year computer science, data science or statistics course.
Likely Majors: PUBHLTH, IOE, KINES
Additional Desired Skills/Knowledge/Experience
If you have any of these characteristics, highlight them on your Experience and Interest Form and talk about them in your (optional) one way video interview.
- Practical design/build experience based in sensors, hardware, and image processing.
- Experience/interest with occupational safety regulations (e.g., OSHA).
- Experience with video processing.
- Demonstrated success in team-based engineering projects (ideally from outside a class)
- Demonstrated leadership experience from any organization.
- Studying construction management (in addition to one of the technical roles above) a strong plus.
- Prior construction experience or military service a plus.
Sponsor Mentor
Anthony Offak
Technology Specialist, Walbridge Innovation and Improvement Leadship
With nearly 25 years of experience, Anthony leverages his passion for emerging technologies to optimize construction workflows. Current responsibilities include implementing innovative solutions to improve safety and efficiency on construction sites. He is continuous improvement focused, and lives by the motto of trying to leave people, places, and things better than when he was introduced to them.
Executive Mentor
Marc Glaeser
Director of Technology, Walbridge Innovation and Improvement Leadership
With 12+ years in the construction industry, including 7+ years on various jobsites, Marc brings a field-first perspective to new technology at Walbridge. He is passionate about finding more efficient, safer, and more effective ways of working on complex construction projects. Current responsibilities include investigating and implementing new technologies and coordinating across various departments to streamline processes and the flow of information.
Eric Twigg
Senior Vice President, Walbridge Innovation and Improvement Leadership
Over his 42 years with Walbridge, if there’s a department at Walbridge that Eric Twigg hasn’t run, it hasn’t been created yet – and Eric will probably be the person who starts it. Throughout his career, he’s led (and often created) virtually every department at Walbridge. Current responsibilities include leading the teams for: New Technology, Innovation, DEI, Quality/Lean, Training, Strategy and Leadership Planning support for all Business Units.
Faculty Mentor
Carol Menassa
Associate Professor and John L. Tishman CM Faculty Scholar, Civil Engineering
Professor Menassa’s research focuses on understanding and modeling the interconnections between the human and the built environment. From her website: In this context, my research group focuses on two main research thrusts. In the first, we study the impact of human behaviors and actions on the built environment. For example, we use modeling and simulation to understand the impact of occupants on energy use in buildings and develop decision frameworks to sustainably retrofit existing buildings. In the second thrust, we focus on understanding the effect of the built environment on human comfort, well-being and accessibility issues. For example, we use non-intrusive methods such as low cost thermal cameras to provide personalized thermal comfort settings in single and multi-occupancy space. We also develop personalized localization and path planning methods to assist people with physical disabilities in navigating unknown building environments. My research group has expertise in energy simulation, complex adaptive systems modeling, high-level architecture and informatics, computer vision and robotics
Weekly Meetings: During the winter 2025 semester, the Walbridge team will meet on Tuesdays from 4:30 – 6:30 PM. Location TBD.
Work Location: Weekly work will take place on U-M’s North Campus. Students will work with Walbridge construction project staff to better understand the user needs. Students will make field visits throughout the project, including to current North Campus building projects, to view current equipment, develop user requirements, and validate their system. (MDP will provide transportation)
Course Substitions: CE MDE, ChE Elective, CS Capstone/MDE, Data Science MDE/Capstone, DS Graduate Practicum, EE MDE, CoE Honors, ROB 490, SI Elective/Cognate
Citizenship Requirements: This project is open to all students. Note: International students on an F-1 visa will be required to declare part time CPT during Winter 2025 and Fall 2025 terms.
IP/NDA: Students will sign standard University of Michigan IP/NDA documents.
Summer Project Activities: No summer activity is planned for the project.
Learn more about the expectations for this type of MDP project