Introduction: The research of robot capture involves a lot of aspects, include machinery, control, the computer, artificial intelligence is waited a moment. A lot of moment, robot capture is an example of very good the least research (Minimal Example) , will support the research of each direction. One, what is robot capture? Following plan institute are shown, give an object and hand, the fundamental problem cent of robot capture is 3: How · catchs · how to go controlling · how to go operating according to difficulty, these 3 problems are main it is ordinal those who increase by degrees. 1, how to go catching? (Grasp Planning) this problem is main it is the issue that a flock of big oxen study at the beginning, include Salisbury, mason, cutkosky, khatib is waited a moment. Everybody considers to consider to go, want to know namely: Give an object, give a job, give a hand, it is best that this hand should catch this content types or forms of literature? The famousest result of this direction is Force - Closure, the importance that is equivalent to the stability inside control basically in a long time. Before I am reading a doctor two years, basic also be to do this, the postdoctoral Sahar that takes me at that time is done in this direction perfectly, nevertheless she turns to get finance all right. This direction goes to what be united in wedlock with Learning basically to cover a road to go up namely recently, the job of OK and referenced Sergey Levine when Google. 2, how to go controlling? (Grasp Control) this problem studies force accuses namely, include a hand to point to endmost force to accuse, tactile control, stiffness control, impedance control is waited a moment. Very long period of time, everybody is in try to calculate what kind of finger capture force just is best (when the word says to begin to read a Master in those days, the first mails to me the stiffness pilot that the Imin Kao of paper paper is taught even if becomes capture) . The job with Martin Buss and group of Mr. Li Zexiang is most famous inside this, nonlinear optimize a problem to be changed into problem of inequality of a linear matrix, can get basically optimizing a result in left and right sides of a few millisecond. The Object that the first-rate job of this the nearest side should be DLR come out - Level Impedance Control (IJRR) (article Wimbock of the first author also turns went) . Still stand fast now in this direction basically is professor of a few Japan, include my co-worker Kenji. The word says the boss Arimoto of Kenji is taught, the ability after retiring begins to do robot capture, still gave this book. He is belonged to antediluvian great mind level, very few now somebody knows, IROS 2016 his birthday party also comparatives high end, can experience below (2006 International Symposium On Advanced Robotics And Machine Intelligence) . 3, how to go operating? (Dexterous Manipulation) among them the Dexterous Manipulation of the last problem (address operation) , had done not have the progress with good what a lot of years, very few also now someone goes meeting this problem. The doctoral paper of Cutkosky is to do this, still did a few years later, changed lab name later again, be turn direction. Have before Peter Allen group postdoctoral do this, be like later is phreatic water was hanged, it is regretful very. The closest Soft Robotics fire, this direction seemed to have new a favourable turn, oliver Broc also begins to make this way. 2, why is robot capture important? Following plan institute are shown, the research of robot capture involves a lot of aspects, include machinery, control, the computer, artificial intelligence is waited a moment. A lot of moment, robot capture is an example of very good the least research (Minimal Example) , will support the research of each direction. Next, the research content in robot capture, especially address sex and alternant sex, can be opposite a lot of otherer action reachs since relevant research, the 2nd piece of graph below showed this. Additional, capture also is the robot walks into real world's necessary function, the hand is so important to the mankind, we hope to gift of course the function with same robot. 3, why is robot capture very difficult? Capture is so simple us to, so we think pair of robots are very simple also very likely, but it is quite difficult actually. If had looked all sorts of perhaps had played all sorts of capture games, should be unripe without can love, the affirmation that the bottom line of robot capture imagines than you wants low. But the data in a lot of Paper is to be below the very truthless requirement that make, give a person the sense of a kind of very high successful rate. Remembered me giving a 30 ％ successful rate of the left and right sides inside my doctor paper, have rejoin the committee member asks me why so low, this does not accord with constant manage. But this is reality. Real world has too much uncertainty, because of uncertainty, the model that we use in capture basically is inaccurate, it is wrong even. And we are done not have again enough good sensor, can give us the state with actual feedback in real time. Worse is, our bit finer even hand is trashy also, control the state that robot hand wants to us very hard well and truly. Future is very long period of time, how to handle these uncertainty, will be the direction of a heat of robot capture. This direction is very actually hot also, basic and annual have relevant Workshop. Promotion issues us here the ICRA Workshop 2017. 4, prospective direction? Since still have the problem that did not solve of so much, so where is prospective direction? Myself summed up a few to nod: Better understanding uncertainty, more using contact, more address design, stabler sensor. Of these a few direction elaborate, also be a very large hole, in the future each get more. (after my doctor graduates, if go to the United States doing postdoctoral, study namely among them the a bitth and at 2 o'clock) . If these a few are nodded,had been done, the reality application that robot capture accumulates from bedding face also won't too far. (article author Li Miao, wuhan Cooper is special author of science and technology, doctor of industry of grain of federal of mulberry of Swiss the name of a river in Shaanxi and Henan provinces, research direction is coordinated for robot study and man-machine alternant. ) CNC MillingCNC Machining