不僅(jin)僅(jin)理解行(xing)(xing)為(wei)(wei),更(geng)需(xu)要(yao)能執(zhi)行(xing)(xing)行(xing)(xing)為(wei)(wei),機器人(ren)能夠執(zhi)行(xing)(xing)行(xing)(xing)為(wei)(wei)才是(shi)真正理解行(xing)(xing)為(wei)(wei)。 通過(guo)計算機視覺以(yi)及行(xing)(xing)為(wei)(wei)認知識別, 讓機器人(ren)能夠確認和(he)分(fen)辨(bian)一個 行(xing)(xing)為(wei)(wei)僅(jin)僅(jin)只是(shi)第一步,這也只是(shi)傳統旁觀式(shi) AI 學(xue)(xue)習所達(da)到的功能水平, 例如,傳統 AI 學(xue)(xue)習可以(yi)讓機器學(xue)(xue)習"盒(he)(he)子(zi),概念并在新的場(chang)景中說出 "盒(he)(he)子(zi), 這個標簽, 但在具(ju)身(shen)智能學(xue)(xue)習模式(shi)中, 機器通過(guo)感知環境實(shi)體(ti), 通過(guo)親(qin)身(shen)體(ti)驗完成(cheng)具(ju)身(shen)學(xue)(xue)習,最終理解場(chang)景并形成(cheng)"打開,這個概念。 因(yin)此(ci), 當機器可以(yi)執(zhi)行(xing)(xing)該行(xing)(xing)為(wei)(wei)才是(shi)具(ju)身(shen)智能的落(luo)腳點(dian)。
執行(xing)(xing)(xing)行(xing)(xing)(xing)為(wei)(wei)需(xu)要(yao)涉及到形(xing)(xing)(xing)(xing)態、行(xing)(xing)(xing)為(wei)(wei)和學(xue)(xue)(xue)習(xi)(xi)的(de)(de)體(ti)(ti)系化(hua)交互。在基于形(xing)(xing)(xing)(xing)態的(de)(de)具(ju) 身(shen)(shen)智(zhi)(zhi)能中(zhong), 形(xing)(xing)(xing)(xing)態、行(xing)(xing)(xing)為(wei)(wei)與學(xue)(xue)(xue)習(xi)(xi)之間(jian)密的(de)(de)切關。系首先(xian),需(xu)要(yao)利(li)(li)用(yong)(yong)(yong)(yong)形(xing)(xing)(xing)(xing)態產 生行(xing)(xing)(xing)為(wei)(wei), 該過(guo)程重點(dian)強(qiang)調(diao)(diao)(diao)利(li)(li)用(yong)(yong)(yong)(yong)具(ju)身(shen)(shen)智(zhi)(zhi)能體(ti)(ti)的(de)(de)形(xing)(xing)(xing)(xing)態特(te)性巧妙地實(shi)(shi)現(xian)(xian)(xian)特(te)定的(de)(de) 行(xing)(xing)(xing)為(wei)(wei), 從而達到部分(fen)取(qu)(qu)代(dai)"計算,的(de)(de)目(mu)的(de)(de)。其次(ci),需(xu)要(yao)利(li)(li)用(yong)(yong)(yong)(yong)行(xing)(xing)(xing)為(wei)(wei)實(shi)(shi)現(xian)(xian)(xian)學(xue)(xue)(xue) 習(xi)(xi), 重點(dian)強(qiang)調(diao)(diao)(diao)利(li)(li)用(yong)(yong)(yong)(yong)具(ju)身(shen)(shen)智(zhi)(zhi)能體(ti)(ti)的(de)(de)探索、操作等行(xing)(xing)(xing)為(wei)(wei)能力主(zhu)動獲取(qu)(qu)學(xue)(xue)(xue)習(xi)(xi)樣(yang)本、標(biao)注信(xin)息(xi), 從而達到自主(zhu)學(xue)(xue)(xue)習(xi)(xi)的(de)(de)目(mu)的(de)(de),此領域當前屬(shu)于研(yan)究前沿(yan)。 再者(zhe), 需(xu)要(yao)強(qiang)調(diao)(diao)(diao)利(li)(li)用(yong)(yong)(yong)(yong)學(xue)(xue)(xue)習(xi)(xi)提(ti)(ti)升行(xing)(xing)(xing)為(wei)(wei)以(yi)及利(li)(li)用(yong)(yong)(yong)(yong)行(xing)(xing)(xing)為(wei)(wei)控(kong)制形(xing)(xing)(xing)(xing)態, 后(hou)者(zhe)有多種 實(shi)(shi)現(xian)(xian)(xian)方(fang)法, 但當前利(li)(li)用(yong)(yong)(yong)(yong)學(xue)(xue)(xue)習(xi)(xi)手段來(lai)提(ti)(ti)升行(xing)(xing)(xing)為(wei)(wei),并進(jin)而控(kong)制形(xing)(xing)(xing)(xing)態的(de)(de)工(gong)作是 現(xian)(xian)(xian)代(dai)人工(gong)智(zhi)(zhi)能技(ji)術發展(zhan)起來(lai)后(hou)涌現(xian)(xian)(xian)出(chu)來(lai)的(de)(de)新型(xing)智(zhi)(zhi)能控(kong)制方(fang)法, 特(te)別是基 于強(qiang)化(hua)學(xue)(xue)(xue)習(xi)(xi)的(de)(de)技(ji)術已成為(wei)(wei)當前的(de)(de)熱點(dian)手段。最(zui)后(hou),具(ju)身(shen)(shen)智(zhi)(zhi)能需(xu)要(yao)利(li)(li)用(yong)(yong)(yong)(yong)學(xue)(xue)(xue) 習(xi)(xi)來(lai)優(you)化(hua)形(xing)(xing)(xing)(xing)態, 強(qiang)調(diao)(diao)(diao)利(li)(li)用(yong)(yong)(yong)(yong)先(xian)進(jin)的(de)(de)學(xue)(xue)(xue)習(xi)(xi)優(you)化(hua)技(ji)術實(shi)(shi)現(xian)(xian)(xian)對具(ju)身(shen)(shen)智(zhi)(zhi)能體(ti)(ti)的(de)(de)形(xing)(xing)(xing)(xing)態 優(you)化(hua)設計。
智能服務機器人 導覽機器人 |