I-algorithm yokudibanisa inzwa

Isihluzo seKalman

Isihluzo se-Kalman siqhelekileyo.

Undoqo we-algorithm kukuseta "inkolelo" imiba kwinzwa nganye. Ngomzuzu ngamnye, idatha yenzwa ukusuka kumzuzu wokugqibela iya kusetyenziselwa izibalo zokuphucula ukuqikelela (ukuzongeza ngokwakho), kunye nomgangatho wenzwa uya kugwetywa. Xa kuthelekiswa phakathi kwexabiso eliqikelelweyo kunye nexabiso elilinganisiweyo lenzwa, ixabiso elihle kakhulu liya kuqikelelwa kunye nesiphumo.

Oku kuthetha ukuba ukuba i-sensor ihlala inika ixabiso elihle kwaye lihambelana kwaye iqala ukukuxelela into engenakwenzeka, inqanaba lokuthembeka kwenzwa liya kuncipha kwii-milliseconds ezimbalwa de iqale ukwenza ingqiqo kwakhona.

Oku kungcono kunomndilili ophakathi okanye ukuvota, kuba i-Kalman filter inokujongana nemeko apho uninzi lwabenzi boluvo lungekho cwangco okwethutyana. Ngethuba nje umntu enokugcina isizathu esifanelekileyo, inokwenza irobhothi ngexesha elimnyama.

Isihluzo se-Kalman sisicelo seengqikelelo ezithe chatha zeMarkov chain kunye neBayesian yokuqiqa, eyinkqubo yemathematika ephinda iphucula uqikelelo lwabo ngokusebenzisa ubungqina. Ezi zixhobo zizixhobo ezisetyenziselwa ukunceda isayensi ngokwayo ivavanye imibono (ekwasisiseko sento esiyibiza ngokuba "ngokubaluleka kwamanani").

Ke ngoko, kunokuthiwa ngokwembongo ukuba ezinye iinkqubo zokudityaniswa kwenzwa zivakalisa eyona nto iphambili yenzululwazi ngesantya sezihlandlo ezili-1000 ngomzuzwana.

Izihluzi zeKalman bezisetyenziswa kwizikhululo ze-orbital zesathelayithi zasemkhathini amashumi eminyaka. Kuba ii-microcontrollers zanamhlanje zinokuqhuba i-algorithm ngexesha lokwenyani, ziya zithandwa ngakumbi kwiirobhothi.