Loading...
Please wait, while we are loading the content...
Similar Documents
Using Global Appearance Descriptors to Solve Topological Visual SLAM
| Content Provider | Semantic Scholar |
|---|---|
| Author | Rojo, Lorenzo Fernández Payá, Luis Amorós, Francisco Reinoso, Oscar |
| Copyright Year | 2018 |
| Abstract | Nowadays, the use of mobile robots has extended to many different environments, where they have to move autonomously to fulfill an assigned task. With this aim, it is necessary that the robot builds a model of the environment and estimates its position using this model. These two problems are often faced simultaneously. This process is known as SLAM (Simultaneous Localization and Mapping) and is very common since when a robot begins moving in a previously unknown environment it must start generating a model from the scratch while it estimates its position simultaneously. To carry out the SLAM process, the robot has to make use of the different sensors it may be equipped with (e.g. odometry, touch, laser, cameras, etc.). During the last years, the use of cameras has extended thanks to the amount of information they can capture from the environment and their relatively low cost. This work is focused on the use of computer vision. When a vision sensor is used to solve the mapping and localization tasks, we must take into account that scenes contain a great quantity of information. This way, it is necessary to extract the most relevant information from the scenes as it will allow us to work with a lower number of components. This problem can be approach from two points of view: local and global appearance methods. First, the methods based on local appearance consist in extracting and describing prominent landmarks or regions from the scenes. These methods typically need more computational time to build the map and estimate the position of the robot since it is necessary to extract the distinctive features from each image, describe them and make a complete comparison with the data stored in the map. Second, globalappearance methods describe each scene with a unique descriptor that contains information of the whole appearance. These methods tend to be computationally more efficient. The SLAM problem can be approached from three different points of view: metric, topological and hybrid metric-topological SLAM. First, the metric approach consists in representing the position of some landmarks in the environment with geometric accuracy, with respect to a reference system. Using this approach, the position of the robot can be estimated with respect to this system. However, these methods often present a high computational cost. Second, the topological approach represents typically the environment as a graph, where the nodes represent prominent localizations (e.g. rooms) and links are the connectivity Lorenzo Fernández Rojo Miguel Hernandez University, Spain |
| Starting Page | 1127 |
| Ending Page | 1140 |
| Page Count | 14 |
| File Format | PDF HTM / HTML |
| DOI | 10.4018/978-1-5225-2255-3.ch597 |
| Alternate Webpage(s) | https://www.igi-global.com/viewtitlesample.aspx?id=184385&ptid=173015&t=using+global+appearance+descriptors+to+solve+topological+visual+slam |
| Alternate Webpage(s) | https://www.igi-global.com/viewtitlesample.aspx?id=213202&ptid=206309&t=using+global+appearance+descriptors+to+solve+topological+visual+slam |
| Alternate Webpage(s) | https://doi.org/10.4018/978-1-5225-2255-3.ch597 |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
