The conditions of visibility of a user moving through an outdoor or indoor environment affect his safety. In both environments it is fundamental to guarantee a sufficient visibility in order to avoid possible collisions. In the specific case of urban roads the available sight distance (ASD) is that part of the roadway ahead which is visible to the driver and should be of sufficient length to allow a vehicle to stop before colliding with a stationary object. Unfortunately, the estimation of the ASD is still carried out on 2D maps or digital terrain models (DTM). Furthermore, it is very complicated to estimate the ASD on urban roads due to difficulties in measuring the effects of individual sight obstructions in a context where their density and dynamicity is relatively high. It is obvious that a 3D representation of the infrastructure may offer more flexibility when modeling sight obstructions and in capturing the complexity of the urban context. Different consideration must to be made for indoor environment, in which, in particular for public buildings, it is important that the main escape routes are correctly signalized. In these situations, a large number of people must easily flow even in highly articulated environments. It is necessary to identify the easiest paths to follow in order to ensure a fast and efficient escape route. The ways in which a person is able to move within the environment are strongly influenced by the space he is able to see in front of himself. The indoor geometry of buildings and the position of some objects can limits the visibility, affecting the perception of the paths through which it is possible to navigate. Different methods have been analyzed to identify the visible space from a certain point of view, by linking the user's perception with the surrounding environment. However, many of these studies are aimed at a two-dimensional environment analysis, or do not take into account a possible dynamic configuration of the motion. The aim of this research work is to propose automatic methods for the visibility analysis, indoor and outdoor, that take into account the dynamism of the individual and the environment. Since it is essential to have a 3D model on which to base the analyses, the first part of the research, will be focused on the investigation of rapid and flexible survey methods, capable of adapting to different contexts of interest. Today the geomatics field already offers different solutions and technologies for collecting 3D information of objects or environments at different scales. In recent years, the development of mobile mapping systems (MMS) that integrate active, passive and positioning sensors, has made a substantial increase in applications in the close-range domain. Moreover, the growth of software solutions for the extraction of point clouds from non-metric image sets has obtained great attention; in fact, these systems combine a good quality of results, both from a metric and a qualitative point of view, simplicity in their use and their accessibility, thanks also to the various free and open source solutions. The main issue regarding the existing MMSs is the high cost of these technologies, which limits their use. A goal of this work is related to evaluate the possibility to realize and to use low-cost MMSs as an alternative for obtaining 3D information of urban infrastructure. Here the term MMS is used in its broadest meaning, considering MMSs all those mapping systems that can be mounted on any type of mobile platform (aerial or terrestrial). In particular, low-cost MMSs mounted on a vehicle or bicycle or carried out by a pedestrian (such as backpacks) will be proposed. Today the mass-market offers a wide choice of low-cost sensors for image acquisition and positioning. Among the imaging sensors, it is possible to find webcams, action-cams and other solutions that can capture high-resolution images while maintaining a very small size and weight. However, these sensors are born with different purposes from the photogrammetric one and, therefore, their use must necessarily be preceded by an analysis of their intrinsic characteristics and their performances and limits under different conditions of use. Starting from these assumptions, the characteristics of multiple low-cost imaging sensors (an action-cam Garmin Virb Elite, Logitech HD Pro Webcams C920, Raspberry Pi Camera Modules v2 and a panoramic camera NCTech iSTAR Fusion) and their intrinsic parameters have been evaluated. Subsequently, after the realization of different tools for the management of the sensors, their synchronization in time and the data storage, some tests were performed in an urban environment in order to evaluate the performances of different configurations of low-cost MMSs and the accuracies of the obtainable 3D models. The input data used for the 3D information extraction were video and frames, which were processed through Structure from Motion (Sfm) technique. The second part of the activities aimed at the development of automatic procedures to perform visibility analysis exploiting the obtained 3D model of the environment. Different tools for the visibility analysis and the estimation of the ASD from a driver’s point of view were compiled and tested. These approaches aimed at identifying all the objects that obstruct the sight of the driver along the road infrastructure. The visibility analyses were performed through ArcGIS and Matlab® tools using the Line of Sight (LoS) approach, and taking into account the dynamicity of the movement. Finally, a procedure for quantitatively assessing the complexity of a known path to follow in emergency cases inside of public buildings was proposed. The methodology combines the use of isovists and visibility graphs techniques and allows to extract numerical descriptors of the environment.
Visibility analyses using 3D urban models generated by low-cost multi-sensor approaches / Grasso, Nives. - (2018 Apr 16).
Visibility analyses using 3D urban models generated by low-cost multi-sensor approaches
GRASSO, NIVES
2018
Abstract
The conditions of visibility of a user moving through an outdoor or indoor environment affect his safety. In both environments it is fundamental to guarantee a sufficient visibility in order to avoid possible collisions. In the specific case of urban roads the available sight distance (ASD) is that part of the roadway ahead which is visible to the driver and should be of sufficient length to allow a vehicle to stop before colliding with a stationary object. Unfortunately, the estimation of the ASD is still carried out on 2D maps or digital terrain models (DTM). Furthermore, it is very complicated to estimate the ASD on urban roads due to difficulties in measuring the effects of individual sight obstructions in a context where their density and dynamicity is relatively high. It is obvious that a 3D representation of the infrastructure may offer more flexibility when modeling sight obstructions and in capturing the complexity of the urban context. Different consideration must to be made for indoor environment, in which, in particular for public buildings, it is important that the main escape routes are correctly signalized. In these situations, a large number of people must easily flow even in highly articulated environments. It is necessary to identify the easiest paths to follow in order to ensure a fast and efficient escape route. The ways in which a person is able to move within the environment are strongly influenced by the space he is able to see in front of himself. The indoor geometry of buildings and the position of some objects can limits the visibility, affecting the perception of the paths through which it is possible to navigate. Different methods have been analyzed to identify the visible space from a certain point of view, by linking the user's perception with the surrounding environment. However, many of these studies are aimed at a two-dimensional environment analysis, or do not take into account a possible dynamic configuration of the motion. The aim of this research work is to propose automatic methods for the visibility analysis, indoor and outdoor, that take into account the dynamism of the individual and the environment. Since it is essential to have a 3D model on which to base the analyses, the first part of the research, will be focused on the investigation of rapid and flexible survey methods, capable of adapting to different contexts of interest. Today the geomatics field already offers different solutions and technologies for collecting 3D information of objects or environments at different scales. In recent years, the development of mobile mapping systems (MMS) that integrate active, passive and positioning sensors, has made a substantial increase in applications in the close-range domain. Moreover, the growth of software solutions for the extraction of point clouds from non-metric image sets has obtained great attention; in fact, these systems combine a good quality of results, both from a metric and a qualitative point of view, simplicity in their use and their accessibility, thanks also to the various free and open source solutions. The main issue regarding the existing MMSs is the high cost of these technologies, which limits their use. A goal of this work is related to evaluate the possibility to realize and to use low-cost MMSs as an alternative for obtaining 3D information of urban infrastructure. Here the term MMS is used in its broadest meaning, considering MMSs all those mapping systems that can be mounted on any type of mobile platform (aerial or terrestrial). In particular, low-cost MMSs mounted on a vehicle or bicycle or carried out by a pedestrian (such as backpacks) will be proposed. Today the mass-market offers a wide choice of low-cost sensors for image acquisition and positioning. Among the imaging sensors, it is possible to find webcams, action-cams and other solutions that can capture high-resolution images while maintaining a very small size and weight. However, these sensors are born with different purposes from the photogrammetric one and, therefore, their use must necessarily be preceded by an analysis of their intrinsic characteristics and their performances and limits under different conditions of use. Starting from these assumptions, the characteristics of multiple low-cost imaging sensors (an action-cam Garmin Virb Elite, Logitech HD Pro Webcams C920, Raspberry Pi Camera Modules v2 and a panoramic camera NCTech iSTAR Fusion) and their intrinsic parameters have been evaluated. Subsequently, after the realization of different tools for the management of the sensors, their synchronization in time and the data storage, some tests were performed in an urban environment in order to evaluate the performances of different configurations of low-cost MMSs and the accuracies of the obtainable 3D models. The input data used for the 3D information extraction were video and frames, which were processed through Structure from Motion (Sfm) technique. The second part of the activities aimed at the development of automatic procedures to perform visibility analysis exploiting the obtained 3D model of the environment. Different tools for the visibility analysis and the estimation of the ASD from a driver’s point of view were compiled and tested. These approaches aimed at identifying all the objects that obstruct the sight of the driver along the road infrastructure. The visibility analyses were performed through ArcGIS and Matlab® tools using the Line of Sight (LoS) approach, and taking into account the dynamicity of the movement. Finally, a procedure for quantitatively assessing the complexity of a known path to follow in emergency cases inside of public buildings was proposed. The methodology combines the use of isovists and visibility graphs techniques and allows to extract numerical descriptors of the environment.Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/11583/2705900
Attenzione
Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo