Real Time Landmark based GPS Navigation Apurva Kumara, Karanam Vineethb Samsung India Mobile Software, South West Asia R&D [Abstract]

1.2 The purpose of your paper

This paper deals with an improved landmark based GPS visual navigation system for mobile devices along with features like path navigator, mapping the city map using landmark based Pedestrian Navigation approach and context based searching.

This paper points out the shortcomings of the existing GPS based navigational models and throws a limelight on the new Real Time Landmark based GPS Navigational system which uses the geotagged images. Also this paper introduces new concepts like visual navigation, mapping the cities, getting the trace of a path, uploading the photos and introduction to context based searching

1. Introduction 1.1 Industry and Technology trends In this modern era, mobile phones have become an integral part of our daily needs and this can be a tool for new location based services. Their ubiquity, portability, good connectivity, the trend towards increased performance and inclusion of new sensors such as GPS and cameras make them an ideal target device. In particular, navigation is emerging as a critical application for the mobile phone industry. Our paper is inspired and motivated by the two papers Visual Navigation for Mobile Devices [1] and LandmarkBased Pedestrian Navigation from Collections of Geotagged Photos [2]. Our work extends a standard mobile navigation system with a pedestrian mode. It uses an existing user contributed database of geotagged photographs to automatically generate navigational instructions using photographs of landmarks augmented with directional information. According to recent research studies using mobile devices [3,4,5], landmark-based navigation instructions are easier to follow, shorten the navigation time, and reduce confusion by providing an instant visual feedback on the correctness of a navigation decision. Landmark-based navigation is particularly effective when the landmarks are clearly visible and the directions are properly embedded in the photographs, and when it is integrated with a standard map-based system. A service that utilizes landmark-based navigation faces the problem of automatically creating and visualizing the instructions. It is generally accepted that approaches such as Google Maps Street View [6] are not well suited for mobile devices, since they are too intensive and frequently suffer from limited landmark visibility [7]. We explore the possibility of leveraging existing collections of geotagged photos acquired using mobile phones for automatically generating navigational instructions. We believe this is a promising new approach, since these databases are more targeted towards useful information and are inherently more scalable.

1.3 Strategic fitness Navigation is one of the most important applications that is going to be vital in the future for mobile devices. The need toward better performance and the inclusion of cameras and GPS sensors make it feasible for developing more compelling location based services. Mobile phones’ ubiquitous connectivity also give users access to online sites, allowing them to both consume and produce geographically meaningful data, such as geotagged images for Flickr.com. Imagine a scenario where a tourist visits an unfamiliar city. Locating oneself on a paper map is often cumbersome because pairing labels on the map with labels in one’s surroundings might be difficult. GPS-based solutions alleviate part of this difficulty, but they still require users to make associations between abstract representations of their environment (maps, turn directions, and so on) and their physical surroundings. This correlation can require significant mental effort. In our day to day life, the use of landmarks is an optimum way for people to navigate and can decrease cognitive load. 1.4 Breakthrough point of the research While using the GPS system, the user has to map the abstract images and data to the real time environment requiring much cognitive work. And considering the limitations of Google Maps Street View, we are using a landmark based approach which facilitates the user to navigate easily using the geotagged images and also we provide context based search to get the details of the nearest landmarks by using the metadata provided to the geotagged images. And even if the user doesn’t get the relevant landmark images, the user can get a 3D view of the surroundings and respective annotations, and directions to proceed from that point.

2. Main Part In this paper, we introduced some new concepts for generating navigational directions and instructions. The main aim of our landmark based navigation approach is to show a recognizable landmark in the environment and to indicate where to go with respect to that landmark. In the previous works, an image of the landmark was rendered with a directional arrow on top of it. This arrow can be oriented correctly by using the reconstructed camera pose.

Figure 1: Algorithm For Real Time Landmark based GPS Navigation

If we want to move from the current location to the nearest Hospital, using the algorithm, we can see that first a GPS map view is generated as shown in the Figure 2. It will give the user the available landmarks available along the path.

Instead of simply piling up the existing database images (like Flickr.com, Panoramio.com), we alter them to produce better views and provide richer context with more realistic augmentations. We used two approaches: one that warps the images from the database to synthesize a camera view aligned with the navigation path, and one that uses a 3D model of the scene synthesized using an automatic technique. In this approach, we use available information about the scene (camera poses and building outlines) to highlight landmarks and provide a user interface for the users to follow a hyperlink of an element of their environment, as seen in Figure 3. Now, in addition to having easier navigation, our tourist can easily access extended information about the current environment. The UI part involves the user entering the final destination which in turn generates a GPS map view containing all the landmarks with respect to the destination. The entire database surrounding the final destination will be searched for the landmarks and available landmarks will be highlighted in the path. Now the user has to compute the camera pose and head towards the nearest landmark and can get the navigation instructions from the landmark with respect to the destination. The Algorithm which is shown in the figure 2 summarizes our approach.

Figure 2: GPS based map view generated containing all the landmarks along the path. (A path from current location to the hospital via University highlighted in red color has been shown)

Once the user reaches one particular landmark, he will get the directions to the next landmark from that current landmark. As we can clearly see from the picture that, the user want to move to the hospital and after reaching the junction, he is encountered with a landmark and he gets the navigational instructions with respect to that landmark. In this case, the user has encountered a university. So he got the directions from the University which says “Take a right turn from the university and keep straight for 50 meters to reach the District Hospital”. The green arrow represents the direction visually (Figure 3).

Figure 3: An example showing navigation and landmark selection.

Whenever we don’t have an image matching the landmark we use the automatically generated 3D models to construct virtual worlds or 3D maps. These can be used to synthesize new views from widely varying viewpoints, which can easily be chosen to include additional context, such as paths and streets. These models are small enough to be rendered on today’s mobile devices, opening up new user interface possibilities. An example of a navigational image generated using this 3D model is shown in Figure 4

Figure 4: An example of a view rendered using 3D models.

another landmark) he can maintain the track of the path which he has travelled by bookmarking the places on the way. Now this entire trace (of the path) can be saved and even send to other GPS users. So that, when another person wants to travel from one particular landmark to another particular landmark, he has got the information from the saved trace and can easily travel using the trace. First we start with the famous cities mapping, i.e. a lot of images of the important places in a particular tourist spot will be uploaded. So that whenever a new tourist comes to the city, he doesn’t have any difficulty in going through the city. In the later stages we can start providing with the mappings of the other cities too. An important feature that can be really helpful is the concept of content based searching. When a person is standing at a particular point, he can search for different places like shopping malls, hospitals, restaurants etc. in and around that location. When he finalizes to travel to a particular location from the existing point he will get a landmark based visual aid to move from his present location to the final destination.

Figure 5: An example of a zoom-out view image. The arrows indicate a path that passes to the right of the building, even though an image from this location is not available in the database. This is the same path as specified in Figure 5.

Also if the user thinks that there is no relevant picture of a particular landmark, we allow the user to upload the photos and information using geotagging. This helps in increasing the database size of the images and image matching will be quite easier. So, whenever a person sees a monument or a landmark, he can take the picture and upload it. It will be helpful to him, when he returns back to the same spot after several days and also it will be also useful to the new tourists who may visit that place for the first time. The land mark is also provided with the hyperlink which can be used to get the additional data about the landmark. Refer Figure 6. Here we are masking the landmark and providing the hyperlink. When the user clicks on the mask, he will be redirected to web search engine results. Another addition is that when a person is moving from one place to another (rather we say it as, one landmark to

Figure 6: This images shows the camera pose image and the same image in which hyperlink is generated after identifying it as a geotagged landmark. Clicking on the hyperlink will provide information about this landmark like Wikipedia links and encyclopedia audio files. 3. Conclusion We have presented the design and evaluation of a prototype system that can automatically generate landmark-based pedestrian navigation directions from existing collections of geotagged photos. Both the selection of navigation images and augmentation of those images with directional instructions are done completely automatically. There are still areas of future work to pursue like adding the voice clips and animations to the navigational instructions. We can also provide extensive results in the

context based search like providing the price range for hotels, restaurants, book stores, medical shops etc. for quotidian usage. 4. Reference [1] Harlan Hile, Alan Liu, and Gaetano Borriello, Radek Grzeszczuk and Ramakrishna Vedantham, Jana Kosˇecka, Visual Navigation for Mobile Devices, IEEE Transactions on April 2010 (vol. 17 no. 2), pp. 16-25 [2]H. Hile et al., ‘‘Landmark-Based Pedestrian Navigation from Collections of Geotagged Photos,’’ Proc. ACM Int’l Conf. Mobile and Ubiquitous Multimedia (MUM), ACM Press, 2008. [3] Beeharee, A.K., Steed, A.: A Natural Way finding Exploiting Photos in Pedestrian Navigation Systems. In: MobileHCI '06: Proceedings of the 8th Conference on Human-Computer Interaction with Mobile Devices and Services, ACM Press (2006) 81- 88 [4] Millonig, A., Schechtner, K.: Developing LandmarkBased Pedestrian-Navigation Systems. Intelligent Transportation Systems, IEEE Transactions on 8 (2007) 43 - 49 [5] Kolbe, T.H.: Augmented Videos and Panoramas for Pedestrian Navigation. In Gartner, G., ed.: Proceedings of the 2nd Symposium on Location Based Services and TeleCartography. (2004) [6] Google: Google Maps Street View. http://maps.google.com/help/maps/streetview/index.html (2007) [7] Walther-Franks, B.: Augmented Reality on Handhelds for Pedestrian Navigation. Master's thesis, University of Bremen (2007)