The Study For Mobile Manipulator About The Detection And Tracking Technology Of Moving Object

The thesis is supported by a National Program 863 item--"The Mobile Manipulator system aimed at measurement and maintenance, working in the circumstance of chemistry dangerous leakage from the chemistry dangerous reactor"(item serial number: 2003AA421040). We adopt the HEBUT-â…¡mobile manipulator as the research object to do some correlative technology research. This paper firstly introduces the hardware system of mobile manipulator, then analysis the machine framework of hand part of mobile manipulator and work space of the whole mobile manipulator. About the kinematics and dynamics of mobile manipulator system, this thesis carries on the relevant theory analysis and calculates, builds up the kinematics and dynamics mode of the whole mobile manipulator system and does relevant simulation. In the vision aspect, we mainly research the image recognition, three-dimension reconstruction of moving target based on double-CCD and vision servo tracking to moving target. The main innovation work of this thesis is as follows:1.In the image disposing and computer vision, a great deal arithmetic adopt HSI color mode, the I parameter of HSI mode is independent of image color information. The H and I parameter is relevant with the color information, and that don't changed along with the light condition, so which is fit to apperceive the color characteristic. This paper adopts HSI color threshold segmentation method only to analysis the H and S parameter value of each pixel of originality image. By this way, we can find the color distributing character of background region and target region. We separarly confirm the rational threshold of H and S parameter to do"and"operation, then separate out the target region from the background region.2.Because the gathered image passes through segmentation and pre-processing, it need a great deal of operation and expend a mass of time. This is disadvantageous to fast detection the moving target. So this paper puts forward the idea of average segmentation, whose basic theory is: after judging out the target region, the two value image of above step obtaining will be divided into n*n part. In each part, if the proportion of the black pixel of the whole target region exceeds a certain threshold, then we consider that this part is background region, until the all part is detected. By this kind of dealing with, the amount of information is reduced. At the same time, in the precondition of guarantee image not distortion, it can lowered the amount of calculation of image processing and realize the image filter. 3. This paper combines the theory and arithmetic of computer vision with the image processing. In the experiment, we use camera screen target, analysis and process the target image, then carry through the three-dimension reconstruct. By this way, we can make sure the scope orientation of target and measure distance. The camera demarcate is necessary to the three-dimension reconstruct of scope object. The method in this paper analyzes and compares common camera model and its calibration, bring forwards a new calibration method, namely the least two multiplication. During calibration, it can reconstruct the object through the projective matrix, avoid solving the parameter of the camera inside and outside, decrease calculation complexity.4. In the process of tracking moving target, for improve the processing speed and satisfy the request of real time, we adopt the parallel processing way, that is when the image gathering card gathers images, computer may continue deal with the data of image. By this way, the gathering image doesn't impropriate the CPU time and can use all time to deal with the data. In consideration of the request of real time and the complexity of arithmetic, in the process of tracking target, we make certain the moving aspect by contrast the change of target mass center position of fore-and-aft image. This way is not only simple, but also practicality. It is easy to realize the fast tracking to moving target.5. We adopt Monto Carlo way, making use of the positive kinematics of manipulator, to draw the work space. Fact shows this way is fastly and stably, which can provide theory foundation and applying guidance for the location of mobile manipulator and work space.6. In the foundation of researching the positive and negative kinematics mathematics theory of mobile manipulator, we establish the whole kinematics mode for mobile manipulator and change the working track of discards into joint space, accordingly accomplish the calculating that manipulator move along the particular track. Then, we adopt the way of Lagrange to establish the dynamics mode for the whole mobile manipulator system. For simplifying this complicated system, we neglect some factor, for example, friction force,disturbing of the ground and so on. Then, respectively research the dynamics of 5-DOF manipulator part and mobile carrier part and establish the dynamics mode for them. Under the environment of UG, we build up the 3-D mode for mobile manipulator, and then transmit it into ADAMS software to do the simulation analysis of virtual prototype. In the end, we get the real time variation situation of joint moment and the simulation curve between joint moment and joint turning angle.