Obstacle Detection in Railway Environment: Filtering the output of an on-board detection system using railway maps
Nisula, Eetu-Veikko (2021)
2021
Automaatiotekniikan DI-ohjelma - Master's Programme in Automation Engineering
Tekniikan ja luonnontieteiden tiedekunta - Faculty of Engineering and Natural Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2021-05-18
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202105104697
https://urn.fi/URN:NBN:fi:tuni-202105104697
Tiivistelmä
The purpose of this thesis was to examine the differences in railways and roads as traffc environments in the context of obstacle detection, and to develop a prototype system that aims to satisfy the requirements of railway environment in varying weather conditions. There were three research questions that considered (1) improving obstacle detection on railways compared to the state of the art, (2) the differences between the aforementioned traffc environments, and (3) if it is possible to detect natural targets reliably from long distance with the developed system. The need for this research is backed by the fact that the upcoming unifying railway infrastructure safety systems European Rail Traffc Management System (ERTMS) and European Train Control System (ETCS) could be improved by using on-board detection systems for detecting various kinds of obstructions that may cause train derailments and other damages.
This thesis includes literature research of roads and railways as different traffc environments emphasizing the differences from obstacle an detection point of view. Additionally, various detection technologies as well as the state of the art were reviewed. On the side of empirical studies, a prototype of the detection system and a data processing algorithm were conceived. An experiment was carried out to get results by keeping eye on the third research question in particular: the system was mounted on the side of a road where detections could be obtained, allowing the evaluation of both the data processing algorithm and hardware performance.
The literature research results indicate that there is defciency in current obstacle detection systems’ reliability in all-weather conditions as well as the detection distance in railways that needs to be covered. Moreover, many automotive and sensor manufacturers along with research institutions have focused mainly on the road environment in the context of on-board obstacle detection systems, whereas railways have lacked the same interest. Because these traffc environments are very distinct, for example noticing the differences in braking distances and in the ability to dodge obstacles let alone vehicles’ differences and the amount of traffc fow, there is a need for sensory equipment and data processing that is especially designed for railway traffc use. The experiments conducted showed that the system is capable of providing detections up to 200 meters, which is not enough to ensure adequate braking distance for an average running train. The results however can be used to guide the design of a more suitable detection system and to pinpoint critical areas regarding railway safety.
This thesis includes literature research of roads and railways as different traffc environments emphasizing the differences from obstacle an detection point of view. Additionally, various detection technologies as well as the state of the art were reviewed. On the side of empirical studies, a prototype of the detection system and a data processing algorithm were conceived. An experiment was carried out to get results by keeping eye on the third research question in particular: the system was mounted on the side of a road where detections could be obtained, allowing the evaluation of both the data processing algorithm and hardware performance.
The literature research results indicate that there is defciency in current obstacle detection systems’ reliability in all-weather conditions as well as the detection distance in railways that needs to be covered. Moreover, many automotive and sensor manufacturers along with research institutions have focused mainly on the road environment in the context of on-board obstacle detection systems, whereas railways have lacked the same interest. Because these traffc environments are very distinct, for example noticing the differences in braking distances and in the ability to dodge obstacles let alone vehicles’ differences and the amount of traffc fow, there is a need for sensory equipment and data processing that is especially designed for railway traffc use. The experiments conducted showed that the system is capable of providing detections up to 200 meters, which is not enough to ensure adequate braking distance for an average running train. The results however can be used to guide the design of a more suitable detection system and to pinpoint critical areas regarding railway safety.