University of Twente Student Theses

Login

Autonomous navigation for the pipeline inspection robot "PIRATE"

Kesteloo, T.V. (2020) Autonomous navigation for the pipeline inspection robot "PIRATE".

Full text not available from this repository.

Full Text Status:Access to this publication is restricted
Embargo date:30 November 2021
Abstract:Smart pipe inspection robots are used more and more to perform quality control from inside pipelines. The "Pipe Inspection Robot for AuTonomous Exploration" PIRATE, developed at the University of Twente, is a snake-like pipe inspection robot that can traverse such pipe networks by clamping itself between the pipe walls of varying pipe diameters and pushing itself along with its wheels. A rotational module allows the PIRATE to align with bends and junctions encountered while traversing the pipe network. This research increases the autonomy of the PIRATE by means of the newly developed control layer that uses the existing PIRATE motion primitives. The control layer is a centralized approach to Path Planning and Motion Planning, responsible for generating a path in a known three dimensional pipe network, and converting the path to the appropriate motion sequences for the PIRATE. Path Planning uses a graph-node based map for the pipe network where each node is an elbow bend or T-junction in the network. These landmarks are connected to one another with straight pipes. A shortest path from any starting landmark to any goal landmark is calculated using the A*-algorithm, using the 3D Euclidean distance between the traversable nodes as cost-function, and the 3D Euclidean distance between the node and the goal as heuristic function. The path consists of motions for a wheeled vehicle with regards to the frame of the pipe network. Motion Planning uses the state of the PIRATE robot to evaluate which move is required next. This converts the motions provided by Path Planning from a wheeled robot perspective to motions that are applicable to the PIRATE. The combination of Path Planning and Motion Planning allow for calculating the quickest path with the A*-algorithm. Using a modified cost-function that penalizes rotating the PIRATE, a path to the goal landmark requiring the least amount of rotations for the PIRATE is calculated. Tests in the simulation environment V-REP show that a digital model of the PIRATE can be instructed to move through a known (mapped) pipe network that consists of challenging sections like vertical pipes. The test uses a simulated corner detection method which allows the PIRATE to detect the subsequent landmarks and their orientation. The PIRATE receives a sequence of motions to perform once a landmark has been detected in order to continue the mission. In case the robot becomes stuck, autonomy can be temporarily deactivated and allow the operator to manually instruct the PIRATE to perform motion primitives. Further research should focus on combining the work on Corner Detection, Feature Extraction, and Mission Control for the physical PIRATE robot, fitting the physical model with the appropriate sensors to accommodate them. Additionally, more fine-grain control is required for the rotational module of the PIRATE to (re-)align for corners.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Embedded Systems MSc (60331)
Link to this item:http://purl.utwente.nl/essays/85373
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page