Cane blanche 2.0

Vocable (Anglais) - - Science -

Les cher­cheurs du MIT ont mis au point un nou­veau sys­tème de na­vi­ga­tion pour les per­sonnes mal­voyantes. Une ca­mé­ra 3D capte l’environnement et un sys­tème de mo­teurs en­voie des vi­bra­tions dans le corps lorsque les obs­tacles se rap­prochent. Dans ce do­maine très concur­ren­tiel, de nom­breux autres la­bo­ra­toires ont dé­jà es­sayé d’amé­lio­rer le dé­pla­ce­ment des aveugles. Ce­lui-ci se­ra-t-il le bon ?

For cen­tu­ries, canes have ser­ved blind and par­tial­ly sigh­ted people well by gi­ving them a means to ne­go­tiate the world around them. The on­ly se­rious up­grade they have un­der­gone dates back to 1921, when a Bri­ton called James Biggs, who had re­cent­ly lost his sight, pain­ted his own cane white in or­der to make it ea­si­ly vi­sible and to alert oth- ers to the pre­sence of so­meone unable to see near­by obs­tacles. In the opi­nion of Da­nie­la Rus of the Mas­sa­chu­setts Ins­ti­tute of Tech­no­lo­gy (MIT), ho­we­ver, the white cane has had its day. Dr Rus would like to re­place it with a sys­tem that scans its user’s en­vi­ron­ment and com­mu­ni­cates back to him what it sees.


2. Dr Rus’s de­vice, of which she de­mons­tra­ted a pro­to­type on June 1st at the In­ter­na­tio­nal Con­fe­rence on Ro­bo­tics and Au­to­ma­tion in Sin­ga­pore, consists of a ca­me­ra worn on a la­nyard around the neck, and a belt. A com­pu­ter in­side the ca­me­ra creates a three-di­men­sio­nal image of the area ahead of the wea­rer, pro­cesses it to ex­tract re­le­vant in­for­ma­tion, and uses the re­sults to pass on ap­pro­priate si­gnals via the belt.

3. Dr Rus knew from pre­vious at­tempts to build de­vices of this sort that what might seem the ob­vious way of ma­ni­fes­ting those si­gnals, na­me­ly as sounds with spe­ci­fic mea­nings, was not, in fact, a good ap­proach. Blind people de­pend a lot on their hea­ring and do not like it when new­fan­gled de­vices ham­per this sense with beeps and clicks. Hence the belt, which has five vi­bra­ting mo­tors ins­tal­led in it. One sits over the centre of the wea­rer’s ab­do­men. The others flank this cen­tral mo­tor, with two spa­ced out on ei­ther side of it.

4. That confi­gu­ra­tion per­mits the com­pu­ter to warn a wea­rer when he is on a col­li­sion course with an obs­tacle. It does so by tel­ling the mo­tor poin­ting most clo­se­ly in the di­rec­tion of the obs­tacle to vi­brate. If the wea­rer is wal­king to­wards a wall, for example, the cen­tral mo­tor vi­brates soft­ly when he comes wi­thin a couple of metres of it. If he ignores this, per­haps be­cause he ac­tual­ly wants to reach the wall, the com­pu­ter in­creases the

am­pli­tude as he closes in, gi­ving him a good idea of exact­ly how far away he is. Si­mi­lar­ly, if he is in dan­ger of bum­ping, say, his right shoul­der on a door­frame while wal­king from one room to ano­ther, the right-most mo­tor on the belt will warn him of the im­pen­ding col­li­sion. And it works. When com­pa­red with na­vi­ga­tion by white cane in one of MIT’s fa­mous­ly crow­ded hall­ways, it re­du­ced blind stu­dents’ col­li­sions with others by 86%.


5. The new sys­tem can, ho­we­ver, do more than just help so­meone walk around wi­thout col­li­sions, for the belt in­cor­po­rates a touch­pad that is ins­cri­bed with ins­truc­tions in Braille. This per­mits the user to pro­gram it to per­form spe­ci­fic tasks. 6. For example, Dr Rus knew that blind stu­dents of­ten struggle to find an empty seat in a crow­ded lec­ture theatre. Ad­ding an ap­pro­priate al­go­rithm to the com­pu­ter’s soft­ware helps get around this by en­abling it to re­co­gnise chairs, and al­so whe­ther or not a chair is oc­cu­pied. In this case, the mo­tors are used to in­di­cate a di­rec­tion to be tra­vel­led in, ra­ther than one to be avoi­ded. Ac­ti­va­ting the al­go­rithm using the touch­pad causes the mo­tor poin­ting most clo­se­ly to­wards an empty chair to vi­brate when the sys­tem spots one.


7. In trials in­vol­ving a room that contai­ned an empty chair, an oc­cu­pied chair and al­so a re­cy­cling bin, the al­go­rithm di­rec­ted the belt-wea­rer straight to the empty chair 80% of the time. Cane users pre­sen­ted with the same ar­ran­ge­ment al­ways found the empty chair even­tual­ly, but in doing so came in­to contact with ob­jects other than their tar­get more than five times as of­ten as those using the ca­me­ra and belt.

8. Whe­ther a ca­me­ra (ideal­ly, smal­ler than the one in the pro­to­type) and a belt could re­place a cane com­ple­te­ly re­mains to be seen. In par­ti­cu­lar, Dr Rus’s sys­tem does lack one im­por­tant fea­ture of Biggs’s in­no­va­tion. A white cane not on­ly helps a blind per­son to na­vi­gate, it al­so si­gnals his condi­tion to the rest of the world, al­lo­wing others to ad­just their be­ha­viour ac­cor­din­gly. As a sup­ple­men­ta­ry aid, ho­we­ver, her ap­proach seems most pro­mi­sing.

The new sys­tem can do more than just help so­meone walk around wi­thout col­li­sions.

Newspapers in French

Newspapers from France

© PressReader. All rights reserved.