ADVANCED FUZZY NETWORKS!!
The performance of the fuzzy-based autopilot system was evaluated by performing a series of approach maneuvering trials using a four-meter FRP boat. As shown in Fig. 12, a CCD camera was installed at the bow of
the boat and connected to a PC running the image-processing scheme and fuzzy autopilot system. The boat was equipped with two outboard motors, one mounted near the bow for berthing purposes, and the other mounted at the stern for propulsion and steering adjustment purposes. During the approach procedure, the distance of the boat from the berthing wall was estimated in accordance with (9) and was also computed directly for reference purposes by a differential global positioning system (DGPS). Finally, the heading deviation angles and tracking deviation angles were computed by the image-processing scheme.
2. Software Structure
As shown in Fig. 13, the software structure was implemented using two PCs, namely a host PC (PC1) and a target PC (PC2). PC1 was interfaced with the CCD via a TV card and executed the image-processing and fuzzy controller schemes utilizing Matlab simulation software with the Simulink
Video and Image Processing Blockset . Meanwhile, PC2, running in the xPC Target environment , was fitted with various A/D, D/A and I/O cards and was used to instruct the steering module of the FRP boat in line with the rudder command signals generated by the fuzzy controller installed on the host PC. The time between the image being acquired by PC1 and the corresponding control command being issued by PC2 was found to be approximately 0.67 sec. Hence, the image processing rate was of the order of 1.5 Hz.