Text

theses

The condition of civil infrastructure such as bridges is of utmost importance for the safety of traveling public and sustainability of the economic activity. The bridge decks deteriorate faster than other bridge components due to their direct exposure to traffic and environmental loads. Effective health monitoring, maintenance, repair, rehabilitation and replacement of the deteriorating civil infrastructure components are necessary to ensure the transportation safety. Current assessment of concrete bridge decks still relies on visual inspection and use of simple nondestructive and destructive evaluations which are not capable to detect defect in early stage. More advanced nondestructive evaluation (NDE) technologies, which can provide more comprehensive assessment, are not used on a regular basis due to lower speed of manual data collection. On the other hand, the current practice of repair of bridge deck only happen at the late stage resulting in extremely high cost. Also, there is currently no available system to treat early stage defect such as delamination and internal cracking.
The goal of this dissertation is to provide a integrated solution for efficient and effective bridge deck inspection and maintenance with emphasis on five interlaced topics: (i) development of an autonomous bridge deck inspection platform, (ii) automated data processing for bridge deck image data, (iii) development of an autonomous bridge deck rehabilitation platform focusing on early stage delamination, (iv) modeling of the bit-concrete interaction for the rehabilitation procedure, (v) strategies for simultaneously deployment of the bridge deck inspection and rehabilitation robots. In the first part, we present a robotic system for bridge deck data collection. The robot integrates multiple NDE techniques that enable the characterization of three most common deterioration types in concrete bridge decks: rebar corrosion, delamination, and concrete degradation. The autonomous navigation and precise data registration are enable by a robust localization system that fusing two GPS and wheel odometry through Extended Kalman Filter (EKF). In the second part, we present a new automated image mosaicing system for bridge deck surface reconstruction. By combining the navigation data and feature-based image registration in the graph optimization framework, our proposed approach inherits the drift-less nature from GPS while still maintains local accuracy of feature-based image registration. In the third part, we develop a robotic system for non-destructive rehabilitation (NDR) targeting the early delamination on bridges such as internal cracking. The NDR system is composed of an omni-directional mobile base, a 5 degree of freedom manipulator and a custom-made end-effector that performs the rehabilitation procedures including drilling and filling. Motion planning algorithm is developed for the mobile manipulator to perform GPS guided rehabilitation procedures.In the fourth part, we present a dynamic model of pure percussive drilling for autonomous robotic rehabilitation for concrete bridge decks. We derive the minimum static force to enable effective percussive drilling which provide us guidance for the mobile manipulator drilling in the previous part. A dry friction-based pure percussive drilling model is then presented to describe the drilling process characteristics and to capture the influence of drilling conditions and parameters on the penetration rate. In the fifth part, we present the strategies to simultaneously deploy the inspection and rehabilitation robot on the bridge decks. We adopt the Gaussian process approach to generate the global and local delamination map online. The inspection robot dynamically determine the step size based on the local prediction uncertainty that accelerate the data collection. Moreover, we design a target planning algorithm based on the global delamination map for the rehabilitation robot to choose the next target to repair. The algorithms proposed are validated through a multi-robot simulation system that could take real bridge inspection data.

ETD_9860

Ph.D.

Includes bibliographical references

doi:10.7282/t3-9cyn-n309

ETD doctoral

The author owns the copyright to this work.