Nondestructive Data Driven Motion Planning for Inspection Robots (AS-5)

Universities: 

University of Nevada, Reno

Missouri University of Science and Technology 

Principal Investigator:

Dr. Hung La, University of Nevada-Reno

PI Contact Information:

Phone: (775) 682-6862  |  Email: hla@unr.edu

Co-Principal Investigators: 

Dr. Genda Chen, Missouri University of Science and Technology                                         

Funding Sources and Amounts Provided:
University of Nevada-Reno: $74,174
INSPIRE UTC: $107,079

Total Project Cost: $181,253

Match Agencies ID or Contract Number:
UNR: In-Kind Match   |    INSPIRE UTC: 00055082-04C

INSPIRE Grant Award Number: 69A3551747126

Start Date: January 1, 2020
End Date: December 31, 2021

Brief Description of Research Project:

According to the Federal Highway Administration (FHWA, 2016), almost one third of 607,380 bridges in the United States are steel bridges. The National Bridge Inventory (NBI) (FHWA, 2016) indicated that 25% of these steel bridges are either deficient or functionally obsolete, indicating a growing threat to the safety of transportation. A number of bridges collapsed recently (e.g., collapse of the I-5 Skagit River Steel Bridge in 2013), strongly suggests for more frequent inspection. Currently, steel bridges need “arms length” inspections and that almost always required the use of the UBITs (man lift). In other cases, human inspectors need to manually climb on the bridge structures to assess condition of the bridges, and it is a difficult and dangerous job. In addition, reports from visual inspection may vary among inspectors so the bridge's condition may not be assessed precisely.

During the first three years, the team led by Dr. Hung La of the Advanced Robotics and Automation (ARA) lab, UNR, has developed four different prototypes of climbing robots, which can climb on steel bridge structure members. These climbing robots would most likely go unseen by the traveling public and could make them a safer alternative for steel bridge inspection. These robotic systems can be equipped with various types of sensors, such as high resolution cameras, and NDE sensors for detection of fatigue cracks, steel corrosion, bolt loosen, etc. The application of this advanced robotic inspection technology will be a benefit for the States and can produce high-quality and accurate condition data, which has the potential to lead to significant long-term cost savings as well. However, these robots are currently controlled manually by the user, and it will be not convenient and efficient for inspection of a large bridge. The user may get difficulty to observe and control the robot. The ARA team plans to develop autonomous localization and motion planning algorithms for these robots to allow them to work autonomously.

This project aims to provide the already built climbing robotic systems an autonomous inspection capability through: development of an accurate and reliable localization algorithm for the climbing robots to allow them to know their position on the steel bridge; development of a motion planning algorithm to allow the robots to safely navigate on the bridge structure members and efficiently inspect the bridge via nondestructive evaluation (NDE) data driven; test and validation of these proposed localization and motion planning algorithms on the robots in both robotics-based simulation environment and real bridges.

Approach and Methodology: Our approach to autonomous localization algorithm will rely on the Maximum Correntropy Criterion Extended Kalman Filter (MCC-EKF), in which MCC-EKF will be designed to fuse various robotic navigation sensors’ data from the global positioning system (GPS), Inertial Measurement Unit (IMU) and wheel encoders to provide reliable localization. MCC-EKF will be able to handle the GPS signal outage (e.g., when the robot climb under steel structures). Once the localization algorithm development is completed, the motion planning algorithm design will rely on the feedback from the NDE sensors. For instance, if the robot sees the surface crack or corrosion through its camera, then it will deploy the Eddy current sensor to perform in depth inspection for these areas. This NDE data driven motion planning approach will save inspection time since it is not necessary to deploy all inspection sensors in all areas.

Overall Objectives: The objectives of this project are to develop autonomous localization and motion planning algorithms to allow the climbing robots to safely and efficiently navigate and inspect the steel structures.The successful development of these two algorithms will allow the robot to operate/inspect the bridge autonomously with minimum user’s intervention and provide a low cost and less traffic disruption inspection solution.

Scope of Work in Year 1: (1) Robot Localization Algorithm Development, (2) Robot Path Planning Algorithm Development, (3) Test and Evaluation.

Describe Implementation of Research Outcomes:
Research outcomes and implementation plan will be described towards the end of this project.

Impacts/Benefits of Implementation:
Impact/Benefits of Implementation will be summarized at the end of this project.

Project Website:
Progress Reports: