The mobile robot can be phenomenologically categorized as the wheeled type, legged type, and the crawl type. Among the these robots of various mobile types, this study covers the wheeled type mobile robot's design, control and free ranging issues. The wheeled type mobile robot is used not only for the automated guided vehicle (AGV), but also for the domestic cleaning, nursing to serve the patient food in the hospital, delivering the documents in the office building, guiding for the blind or the company visitors, and the ultrafine maintenance robot inside the nuclear power plant.
In this study the two mobile type for the developed robot, AGV movement and the translation movement, are explicitly resolved to get the forward/ inverse kinematic and dynamic solutions, which provides capability for the real time calculation as well as the real time dynamics counted control can be possible using the resolved acceleration control method. Simulation results show that the dynamics counted control method is superior to the path tracking method counting the existing kinematic analysis results.
Also, we treat the structure and the principle of the developed omnidirectional laser range finder, induces the simple mapping equations for the range data calculation and the identification of the 3D object's image, image processing algorithm, and the data reduction algorithm for the real time range data processing. Through the experiments, we show that the proposed omnidirectional laser range finder has the high measurement accuracy and the real time identification ability of the environments. The calibration of these laser range finders follows much more complicated process than the usual camera calibration cases, because laser slit emitter must be calibrated as well as the camera itself. CCD camera intrinsic parameters are obtained from the pin-hole camera model. Using these parameters the exact design constants for the laser slit emitter is calculated. The camera location and its rotation is also calculated. Finally the exact 3D range data to the object is obtained utilizing these design constants and CCD camera extrinsic parameters. The calibration method of laser range finder and the calculation algorithm of the exact 3D range data are described.
Finally, this study cover two local path planning algorithms for the mobile robot's free range self-navigation, which are based on the configuration space method and the potential method with the environmental informations and the exact range data obtained from the laser range finder. The local path planning algorithm based on the heuristic method is described in this study, where the robot reaches the target goal following the heuristically accumulated knowledge base which leads the robot to determine go forward, turn, and/or no way out. Also, we deal the local path planning algorithm utilizing the directional weighting method. This method determines the mobile robot's heading direction from the comparison of the attractive resultant force, which is the product of directional weighting function and the range data. And the simple collision-free path and the open pathway test is performed. The heading direction obtained from the above result leads the robot to reach the goal following the shortest pathway without collision to the surrounding obstacles. Also, the feasibility study to the two local path planning algorithms proposed for the robot's real time navigation was performed through the simulation test and experiment under several complicated environments.
