The search is on!
The search is on!Volume 7 Issue 11 News & Resources | November 2014
Even a robot needs to know his position in this world, i.e. location in relation to others. Knowing one’s position will be crucial in the ensuing course of action. With this pertinent information, a robot will be able to navigate its way according to the embedded instructions and ensures the intended mission will be fulfilled. Although, we are fully aware that a robot does not “know” or “aware” of its position in relation to its surrounding to what would be felt by a human being. A robot needs to be supplied of this crucial data (re: position or location), as opposed to a human being who can instinctively implied his or her position from previous experiences, available clues or even a simple act of asking the guy next to him. Localizing one’s position in the world seems to be a very simple and straight-forward activity for us, though, the same situation would not be applicable to a mechanized system, no matter how advanced the so-called “Intelligent System” which we have designed and developed. The process of deciphering natural cues for localization into quantifiable parameters and making sense of them to the robot need more than a casual glance of the problem. It is much stickier than it looks. In the meantime, the robot will just have to make do with whatever information about the world it already has and move forward. The robot cannot solve this problem unless we lend a hand. This is the crux of the matter.
The most important cues for position determination are the natural and man-made objects in the surrounding environments. Some of the cues are static and some are dynamic in nature. Some cues are global, while most are immediate or local. There are also various data form or mode of sensed cues. The level of data complexities are also varied which then led to the reliability of the processed information. A clear example is the quality of cues in ocean localization problem compared to terrestrial data. Ocean localization usually will utilize acoustic information whereas the land-based localization will make use of satellite information via GPS system. Obviously, GPS-based information is more encompassing and accurate compared to those acoustic-type. The primitive parameters in the overall localization determination eco-system will be critical to the accuracy of the final outcome. The margin of error is small. Hence, the reliability of the front end or the sensing side is not less important to the back-end or the processing module. The data fusion and integration of the acquired data will then need to be embedded to the already available map or model of the world. In a way, the acquired data will need to be referenced to the accepted world reference point or axes. At this point, the accepted fact is a consistent enhancement of a reference-map with continuously improved resolution as more data are acquired. More data is certainly good as long as it does not hinder an autonomous system from real-time utilization. And, the localization problems being faced are directly proportional to the level of accuracy needed.
The main difficulty of having a reliable localization system for ocean applications is related to the mode of position detection. The use of sound or acoustic wave led to the difficulties in signal transmission, detection, analysis and display. Acoustic signal, as mechanical wave, is very prone to noise and disturbance. The attenuated acoustic signals at the detection point are also laced with stray signal picked-up along the transmission and/or reflected paths. The actual data is sometimes overwhelmed by the noise portions. The sensor modules configurations play an important role in triangulating the actual position of the sound source. There are three common configurations: Long Base Line (LBL), Short Base Line (SBL) and Ultra Short Base Line (USBL). Depending on the type of applications, sea conditions and access to actual physical localization modules, the appropriate configuration is selected. The major limitation of utilizing the three acoustic-based localization systems is the cost. There are very expensive and typically not suited for small-scaled utilization. Hence, localization becomes a very expensive investment to anyone desiring position information in ocean environment. Localization is a challenge because any visual identification from the surface will not possible for more than 5 meters of water depth. And, typical underwater applications can go down even to 2500m of water depth. Some can go much deeper. Acoustic transponder attached on an AUV becomes the beacon for localization and navigation. The relative location will then be mapped to a Global Positioning System (GPS)-like information, complete with the longitude and latitude. All in all, the localization process in an underwater environment has never been easy and the problems related to it are still yearning to be solved.
Having to locate the position of a single robotic system in an unknown environment is relatively difficult. The task may be prone to multiple complexities due to system’s dynamic, unknown environmental disturbance, and even the more obvious system’s malfunction. What if we are trying to detect multiple robotic systems in that same environment? Surely, the algorithms and processes for localization for such requirement will be far more complex. Then, what if we are trying to locate the locations or positions of different type of robotic systems utilizing different modes of communications? The more pertinent question will not just be “Where am I?” but must be preceded with the question “Who are you?” In reality, this sequence of questions is more relevant and mirrors the real-world applications. The solution of the localization problem will be critical in ensuring the success of the whole application or mission. Of course, an application which involve multiple robotic systems and utilizing multiple communication domains, will give rise to the problem of controller selection and implementation options, and one will have to choose whether a centralized or de-centralized approach. In a nutshell, the issue of identification will need to be settled first.
All system that we built will need some sort of identification(ID) or tag. The tag or ID will be useful for system’s identification in the already crowded real-world environment. The novel controller and navigation algorithms will only be realizable when supplied with the location or position information, absolute or relative. For the time being, the localization problem will continue to be researched on and investigated. Good luck in your search!
|
|