Tham khảo tài liệu 'sustainable rfid solutions part 13', kỹ thuật - công nghệ, cơ khí - chế tạo máy phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | Integration of Data Across Disparate Sensing Systems Over Both Time and Space to Design Smart Environments 289 2. Acoustic and Radio Ranging T ransmissions Broadcast. 3a. Receive Radio Ranging Transmission start timer. 5b. Distance from b to c Reply 4a. Receive Acoustic Ranging Transmission . 3b. Receive Radio Ranging Transmission start timer. 4b. Receive Acoustic Ranging Transmission stop timer. Fig. 5. Acoustic Time-of-Flight Ranging. BS is the base station connected to a computer. The MICA2 motes are labeled a b and c. The fusion of sensor localization and stereo depth map results is performed by a registering the localization and depth map data b estimating the uncertainty of localization and depth map data as a function depth distance and c fusing the two data by minimizing the uncertainty over the entire depth range. The registration problem is approached by either global optimization or local model-based fitting. The global optimization is achieved by minimizing the difference between depth values and localization values in the least squares sense by solving a non-linear set of equations number of MICA sensors is equal to the number of equations using a downhill simplex search. The local model-based fitting approach assumes that a set of apriori known sensors is co-planar. Then the registration is performed by a fitting a 3-D surface to a set of apriori known co-planar sensor locations and b computing the registration transformation parameters. In order to fuse the data the uncertainty of localization and depth map data as a function depth distance is estimated theoretically and verified experimentally. The theoretical uncertainty estimates are derived from a stereo depth disparity equation and from modeling point-point ranging localization error. Figure 6 shows the determination of a fusion threshold for a particular choice localization and depth map uncertainties. We also developed simulation capabilities for any range of input uncertainties to .
