Machining-Based Coverage Path Planning for Automated Structural Inspection

The automation of robotically delivered nondestructive evaluation inspection shares many aims with traditional manufacture machining. This paper presents a new hardware and software system for automated thickness mapping of large-scale areas, with multiple obstacles, by employing computer-aided drawing (CAD)/computer-aided manufacturing (CAM)-inspired path planning to implement control of a novel mobile robotic thickness mapping inspection vehicle. A custom postprocessor provides the necessary translation from CAM numeric code through robotic kinematic control to combine and automate the overall process. The generalized steps to implement this approach for any mobile robotic platform are presented herein and applied, in this instance, to a novel thickness mapping crawler. The inspection capabilities of the system were evaluated on an indoor mock-inspection scenario, within a motion tracking cell, to provide quantitative performance figures for positional accuracy. Multiple thickness defects simulating corrosion features on a steel sample plate were combined with obstacles to be avoided during the inspection. A minimum thickness mapping error of 0.21 mm and a mean path error of 4.41 mm were observed for a 2 m(2) carbon steel sample of 10-mm nominal thickness. The potential of this automated approach has benefits in terms of repeatability of area coverage, obstacle avoidance, and reduced path overlap, all of which directly lead to increased task efficiency and reduced inspection time of large structural assets. Note to Practitioners-Current industrial robotic inspection approaches largely consist of a manual control of robotic platform motion to desired points, with the aim of producing a number of straight scans for larger areas, often spaced meters apart. The structures featuring large surface area and multiple obstacles are routinely inspected with such manual approaches, which are both labor intensive and error prone, and do not guarantee acquisition of full area coverage. The presented system addresses these limitations through a combined hardware and software approach. Core to the operation of the system is a fully wireless, differential drive crawler with integrated active ultrasonic wheel probe, to provide remote thickness mapping. Automation of the path generation algorithms is produced using the commercial CAD/CAM software algorithms, and this paper sets out an adaptable methodology for producing a custom postprocessor to convert the exported G-codes to suitable kinematic commands for mobile robotic platforms. The differential drive crawler is used in this paper to demonstrate the process. This approach has benefits in terms of improved industrial standardization and operational repeatability. The inspection capabilities of the system were documented on an indoor mock-inspection scenario, within a motion tracking cell to provide quantitative performance figures for the approach. Future work is required to integrate the onboard positioning strategies, removing the dependence on global systems, for full automated deployment capability.