University of Twente Student Theses
Lidar and pattern recognition based feature detection in closed gas pipe structures
Vijaykumar, S. van (2021) Lidar and pattern recognition based feature detection in closed gas pipe structures.
|
PDF
5MB |
| Abstract: | A process/system/product has to undergo routine maintenance checks in order to avoid sudden breakdowns and disruption of the process. Currently, maintenance of gas pipelines is a manual, time consuming and expensive process. Robots capable of autonomous navigation has been developed in the past. To use a robot for the application of pipe maintenance it has to autonomously navigate throughout the entire network and successfully identify damages. The PIRATE is being developed into one such robot to conduct autonomous maintenance of gas pipelines. Feature detection is a process which is a necessary step towards achieving this goal. Research has been done into all the feature detection concepts available and two sensory systemconceptswere formulated and tested in this thesis. Here, we consider the combination of a pattern and a camera to detect features which make up sensory system 1 andwe use an array of LIDARs (sensory system 2), using the concept of changes in distance measurements to detect features. The concept of using patterns and LIDARs was tested before but in this thesis we make use of multiple LIDARs and a non-continuous circular pattern (point sources arranged in a circle)in the simulation experiments conducted in order to analyze if these methods are able to overcome some of the constraints faced in earlier conducted experiments. Simulation based experiments were conducted for sensory system 1 and 2. A few parameters were kept constant in both the simulation experiments and the results from both the experiments has been compared, the methodwhich is more adaptable for this project is concluded to be the better sensory system. From the tests conducted in a virtual simulation environment the sensory system using LIDARs provided more accurate results in accordance with the requirements set in Chapter 3. This is further justified when the number of point sources used for setting up the circular pattern are reduced and the number of LIDARs used are also reduced. Both the experiments were tested in the same simulation setup (pipe network) and from the results obtained it is seen that the concept using LIDARs provided the best results. Recommendations for improving feature detection and using these detected features a map of the environment can be built by incorporating an IMU and an Odometer. Another option is the use of LEDs to increase the visibility of the environment which then further helps in using a broader variety of cameras and possibly eliminating the use of patterns. The addition of ultrasonic sensors can be used to determine the thickness of the wall of the pipes which helps in calculating pipe degradation, which is another factor to be checked during pipe maintenance. Further converting the simulated experiments into real-time experiments gives a better idea of the functionality of the sensory system which is another recommendation to be considered. |
| Item Type: | Essay (Master) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Programme: | Electrical Engineering MSc (60353) |
| Link to this item: | https://purl.utwente.nl/essays/104174 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page