Indoor Pedestrian Navigation System Using a Modern Smartphone Alberto Serra
Davide Carboni
Valentina Marotto
CRS4 CRS4 CRS4 Parco Tecnologico, Edificio 1 Parco Tecnologico, Edificio 1 Parco Tecnologico, Edificio 1 Loc. Piscina Manna, Pula, CA, (Italy) Loc. Piscina Manna, Pula, CA, (Italy) Loc. Piscina Manna, Pula, CA, (Italy) +39 070 9250 230 +39 070 9250 303 +39 070 9250 230
[email protected]
[email protected]
communication device with built-in sensors, a wearable camera, an inertial head tracker and display. The method uses the dead reckoning process to detect and measure a unit cycle of walking locomotion and direction achieved by analyzing the acceleration vector, angular velocity and magnetic vector acquired from builtin sensors (in this work is used the 3DM-G from MicroStrain Inc).
ABSTRACT In this work we present a pedestrian navigation system for indoor environments based on the dead reckoning positioning method, 2D barcodes, and data from accelerometers and magnetometers. All the sensing and computing technologies of our solution are available in common smart phones. The need to create indoor navigation systems arises from the inaccessibility of the classic navigation systems, such as GPS, in indoor environments.
The German Aerospace Center studies sensor fusion approaches that combines GNSS (Global Navigation Satellite System), foot mounted inertial sensors, electronic compasses, baro-altimeters, maps and active RFID tags. The system consists of a two-layer sensor fusion architecture that operates with a Kalman filter where possible, and fuses other sensors and maps at a higherlevel, lower rate, particle filter. In buildings, a few dispersed RFID tags can significantly improve the overall performance of the positioning system [2].
Categories and Subject Descriptors H.5.1 [Multimedia information Systems]: Hypertext navigation and maps.
General Terms Measurement, Documentation, Experimentation, Human Factors.
3.
Keywords INTRODUCTION
In this paper, we propose the design and present the early development of an Indoor Navigation System based solely on the capabilities of a modern smartphone equipped with accelerometers, compass, camera and Internet connectivity. Indoor navigation can support commercial activities such as the research of products in a large mall, but can also be deployed for security reasons: evacuation of complex buildings, route identification for visitors etc.
The prototype of this system, as mentioned in the introduction, uses the data from the motion sensors embedded in the smartphone to compute the correct position of the user based on a known initial location. The smartphone application, still under development, is presented in figure 1. The user opens the application and reads with the integrated camera a datamatrix (2D barcode) placed aside the map of the floor (see figure 2).
In the next section we describe the background and some related work in the field of indoor navigation systems. In the third section we introduce the prototype and present the preliminary tests done. The last section draws the conclusions and presents the future steps to be done.
2.
THE MOBILE PROTOTYPE
Differently from the above-mentioned systems, our solution is solely based on the capabilities of a common smartphone. The data read from the phone’s sensors, combined with the reference map of the place, gives the actual position of the user without connecting to any external or pre-installed positioning system such as GPS, RFID, or WiFi trilateration using the dead reckoning technique. Dead reckoning is the process of estimating the current position of a user based upon a previously known position, and advancing that position based upon measured or estimated speeds over elapsed time and course. Errors occurring in the position fix are cumulative, growing with every step the user takes.
Indoor navigation, dead reckoning, accelerometer, compass, map.
1.
[email protected]
RELATED WORK
Based on the URL encoded in the datamatrix, the application downloads from a dedicated server the indoor vector map for the specific floor together with the initial position of the user on the map (corresponding to the point where the user stands when scanning the datamatrix).
In [1] a method of personal positioning for a wearable Augmented Reality system is proposed, allowing a user to freely move around indoors and outdoors. In this system, users are equipped with a Copyright is held by the author/owner(s). MobileHCI 2010 September 7-10, 2010, Lisboa, Portugal. ACM 978-1-60558-835-3/10/09.
397
The step counter module based on the accelerometer data was tested and validated after thorough tests, performed in an indoor environment using both men and women with different physical features. The mean placement error was 3,8% on a series of 20 runs consisting of an average step count of 40 steps. The application based on the compass and the step counter modules, was able to detect accurately both orientation and displacement of an user in an indoor environment, for short runs (less than 100 m).
4. CONCLUSIONS AND FUTURE WORK In this paper we proposed a method for a pedestrian indoor navigation system. We developed this application on a modern smartphone and did first experiments in a real indoor environment, measuring the encountered errors.
Figure 1. Screen of the application with a pedestrian route example.
Future work will include the improvement of the measurement method of the walking steps to overcome the shortcomings of the current-used fixed-value step length. The estimation of the step length could be obtained by the strength of the step acceleration movement (through a probabilistic algorithm) and the personal information data written previously by the user. As seen in the experimental phase, the step counter is subject to accumulated errors, raising the need of a fixing algorithm such for example a particle filter or a Kalman algorithm [3].
When the user starts to walk, the application draws step by step the position of the user, as a continuous line, over the downloaded map of the building floor. The application tracks the number of steps taken by the user based on the data generated by the smartphone’s accelerometers. A single step is detected for each couple of consecutive negative/positive peaks in the acceleration values, i.e. a zerocrossing of the normalized signal generated by the accelerometer. The current orientation of the user is measured by the smartphone’s digital compass (the parameter 'Orientation' in Figure 1). The initial orientation is set when the user scans the 2D barcode, being perpendicular (within a certain angle) to the floor plan. The relative position of the device with respect to the user (e.g. in a pocket) does not influence the dead reckoning estimation. If the device is held in front of the user, the magnetic compass provides the step-by-step heading improving the overall accuracy of the positioning method.
The knowledge of walls, doors, pillars and other elements can be also used for fixing the position errors if these elements are already included as vector data in the floor map, which in this case could be an SVG image. An alternative error correction method might be the use of the integrated smartphone’s camera. The images taken live by the camera could be segmented in order to partition the image into relevant regions. These regions, a simplified representation of the acquired image, can be more meaningful and easy to parse. Using this technique, and using isometric maps of the flat building, it is possible to compare, looking for similar features, the stream of live camera images and the map to get the correct indoor location, fixing the previous accelerometer and compass errors.
5.
REFERENCES
[1] Kougori, M. and Kurata, T. 2003. Personal Positioning based on walking locomotion Analysis with self-contained sensor and a wearable camera. In Proceedings of ISMAR2003, 103-112 [2] Krach, B. and Robertson, P. 2008. Integration of FootMounted Inertial Sensors into a Bayesian Location Estimation Framework. In Proceedings of 5th Workshop on Positioning, Navigation and Communication 2008 (WPNC 2008), Hannover, Germany.
Figure 2. User reads a datamatrix to download the map and his start position.
3.1
Experimental tests and results
[3] Woodman, O. and Harle, R. 2008. Pedestrian Localisation for Indoor Environments. In Proceedings of the 10th International Conference on Ubiquitous Computing (UbiComp), Seoul, Korea, ACM 2008, 114-12
Before starting the application, the compass needs an accurate recalibration. This recalibration is necessary because the compass is subject to several errors: initially it has an inaccuracy of maximum 5 degrees, depending also on the used device and on the presence of electromagnetic interferences.
398
