The application of wearable computers for providing situational awareness in a multi- robot environment. Victor Boskovitz, Alex Pertsev, Hugo Guterman.
The application of wearable computers for providing situational awareness in a multirobot environment Victor Boskovitz, Alex Pertsev, Hugo Guterman Ben-Gurion University 84105 Beer-Sheva, Israel {victor,pertsev,hugo} @ee.bgu.ac.il
Abstract. Recently we are witnessing the adoption of autonomous robotic platforms with cooperation abilities by an increasing number of units in modern armies (IDF among them). Advanced command and control systems are required to operate such platforms efficiently and safely. Furthermore, when used by infantry troops, the command and control hardware should be comfortable to carry and operate “on the move”. Beyond state of the art technologies, the development of such systems requires further research in several fields: wearable computing, mobile access networks and video transmission over specialized networks, power supplies, etc. In this paper we present an experimental system that we developed for the needs mentioned above.
Introduction The most significant amount of innovation, that will enable truly seamless and unobtrusive access to information, is expected from the emergent field of “wearable computing”. The great majority of the published R&D efforts in this field are concentrated on different “core technologies” such as smart textiles, computer miniaturization, relevant sensors, interfaces, etc. Comparatively, there are only a few projects that address the problem of integrating the different components into whole systems. This task is not at all straightforward, because an implementation of the “wearable” concept should be much more than the sum of its parts. Based on our prototype system, we are conducting studies of several technologies in context with this “wearable” concept.
Body Area Network (BAN) Implementation One of our main development issues is the Body Area Network (BAN), required for the communications among the different components of the system. Our implementation (see Fig. 1) is based on the I2C infrastructure and protocol [1], with all components embedded in a special vest (SCOTTEVEST with solar panels [2]). Thus far the network connects the following sensors: temperature, pressure, EKG, GPS and a sensor board (from Crossbow Technology, Inc. [3]) that contains several sensors onboard (acceleration – two axes, magnetic field – two axes, light intensity and a microphone). The interfaces between the sensors and the bus (incl. A/D converters where required) as well as a “gateway” for the computer (iPAQ 5550) where developed in our lab. The network allows immediate and straightforward connection of additional I2C compliant sensors and components. Power for the different components, generated by a combination of batteries and a flexible solar panel, is supplied through a “power network” that allows further connections at several locations on the vest.
Fig. 1. BAN – system components. During normal usage the components are concealed inside pockets and textile channels Applications that employ the data acquired from the sensors for the monitoring of the soldier’s activity and well-being are under development. One application that augments positioning data from the GPS with data from the inertial sensors already shows promising preliminary results: distance measurements based on the accelerometers alone (where GPS signal is unavailable) are under 10%, with tested walking distances of up to 120 meters
on flat ground; acceleration data is also successfully used to detect static states (standing), overriding noisy GPS data.
Inter-Soldier Communications Another research issue of interest relates to the way several wearable systems communicate among themselves, or what we call Inter-Soldier Communications. Special emphasis was put on flexibility and survivability considerations (transparent joining/cutoff of users, non-dependence on a central server, etc.). We evaluated several feasible approaches for implementing such an ad-hoc network while considering the constraints of expected usage scenarios and of the given resources (both WiFi and Bluetooth infrastructures available in the employed PDA’s were evaluated). The implemented network, based on some principles published by the Mobile Adhoc NETworks (MANET) Work Group [4], supports transmission of different message types ranging from text, through audio to real-time video. Video multicast from any user to other users is enabled on the limited available WiFi infrastructure by adapting basic communication protocols to this mission (a combination of the TCP and UDP protocols).
Hierarchical Graphical User Interface (GUI) The overall capabilities of the system are demonstrated by means of a prototype command and control application. The prototype is based on a hierarchical GUI designed to display rich BAN and Inter-Soldier information in an intuitive way on small displays (see Fig. 2).
Applications of Automatic Region Of Interest (ROI) Detection In the framework of other research ongoing in out laboratory we are investigating techniques of automatically detecting Regions Of Interest (ROIs) in images. These are regions that are more informative than the rest of an image, in some sense relevant for the application at hand. Several computational models that attempt to model true behavior of the human visual system were published (e.g. [5]). We have identified three different issues related to wearable computing platforms that can greatly benefit by the successful realization of such an algorithm: 1) Better exploitation of the small displays typical to wearable systems by displaying only ROIs instead of reducing the size of an input image (see Fig. 3).
2) Reducing power consumption by spatially varying the display intensity (feasibility depends on the specific display technology used) and colormap according to the importance of the different image regions (see [6]). 3) Improving the compression ratio of video signals: preliminary results show an average increase of 20% in the compression ratio of video signals for the Inter Soldier Communications module by spatially varying the target quality of modified JPEG and MPEG4 encoders.
Fig. 2. Screen-shots of several displays available in the command and control application’s hierarchical GUI
Fig. 3. Better exploitation of small displays: in some application cropping a large image (top) around a ROI to fit a small display (bottom left) might be preferable over reducing the size of the whole image (bottom right)
References 1. Philips Semiconductors: The I2C-Bus Specification, Ver 2.1, Jan. 2000. 2. SCOTTEVEST Jacket with Solar Panels, [online]: http://www.scottevest.com/v3_store/access_solar.shtml 3. Crossbow Technology Inc.: Motes, Smart Dust Sensors, Wireless Sensor Networks, [online]: http://www.xbow.com/Products/productsdetails.aspx?sid=3
4.
Mobile Ad-hoc NETworks (MANET), [online]: http://www.ietf.org/html.charters/manet-charter.html 5. C. M. Privitera and L. W. Stark, "Algorithms for defining visual regions-ofinterest: Comparison with eye fixations," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, no. 9, pp. 970-982, Sept.2000. 6. Ranganathan, P., Geelhoed, E., Manahan, M., Nicholas, K.: Energy-Aware User Interfaces and Energy-Adaptive Displays, in IEEE Computer, pp. 3138, Mar. 2006.
