抄録
The indoor wireless positioning system for a mobile robot employing fingerprinting technique is made in two phase, offline, i.e., collecting reference data for database, and online, i.e., matching the unknown data to those in the database. It is commonly known that the positioning accuracy is increased with the larger number of the reference locations. This has made the offline phase a tedious works, where laborious efforts are needed to construct the database. Thus, automatic database construction is desired in order to minimize the human efforts. This paper described the Signal Propagated Modified Shepard’s Method (SP-MSM) to construct the database by means of interpolating the missing wireless data using for mobile robot application. By introducing the selection probability, reference locations are identified and database is constructed. We found that over all 64 test locations, the proposed SP-MSM method outperform other interpolation method at 52% locations. In addition, the usage of low pass filter has greatly suppressed the fluctuation problem caused by unpredicted behavior of the wireless signal.
