評(píng)論
打印
Science and Technology
科技版塊
Unimals
通用動(dòng)物
Balls, sticks and the Baldwin effect
球形，棍形和鮑德溫效應(yīng)
A novel way to optimise robots
一種優(yōu)化機(jī)器人的新奇方法
It might sound obvious that if you want to improve a robot's software, you should improve its software.
如果你想改進(jìn)一個(gè)機(jī)器人的軟件，你就應(yīng)該改進(jìn)它的軟件，聽(tīng)起來(lái)像是明擺著的事。
Agrim Gupta of Stanford University, however, begs to differ.
但斯坦福大學(xué)的阿格里姆·古普塔卻不這么認(rèn)為。
He thinks you can also improve a robot's software by improving its hardware—that is, by letting the hardware adapt itself to the software’s capabilities.
他認(rèn)為還可以通過(guò)改進(jìn)硬件來(lái)改進(jìn)機(jī)器人的軟件，這種對(duì)硬件的改進(jìn)是讓硬件去自適應(yīng)軟件的功能。
As they describe in Nature Communications, he and his colleagues have devised a way of testing this idea.
據(jù)他和同事在《自然通訊》中的描述，他們?cè)O(shè)計(jì)出了一種方法來(lái)測(cè)試這一想法。
In doing so, they have brought to robotics the principles of evolution by natural selection.
在此過(guò)程中，他們把通過(guò)自然選擇來(lái)實(shí)現(xiàn)進(jìn)化的原理引入了機(jī)器人科學(xué)領(lǐng)域。
They also cast the spotlight on an evolutionary idea that dates from the 1890s, but which has hitherto proved hard to demonstrate.
一個(gè)有關(guān)進(jìn)化的觀點(diǎn)也因他們而受到關(guān)注，這個(gè)觀點(diǎn)可以追溯到19世紀(jì)90年代，但直到今天都還很難被證明。
There is a wrinkle.
他們想到了一個(gè)好主意。
The team's robots, which they dub "unimals", are not things of metal and plastic.
這支團(tuán)隊(duì)的機(jī)器人——取名“unimal”（通用動(dòng)物）——不是用金屬和塑料制成的。
Rather, they are software entities that interact with a virtual environment in the way that metal-and-plastic devices might interact with a real one.
機(jī)器人都是軟件實(shí)體，可以與虛擬環(huán)境交互，就像用金屬和塑料制成的設(shè)備可以與真實(shí)環(huán)境交互一樣。
Unimals are pretty simple, having spheres for heads and cylinders for limbs.
unimal的模樣非常簡(jiǎn)單，頭是球體，胳膊或腿是圓柱體。
The environments through which they roamed were also simple, and came in three varieties: flat arenas, arenas filled with hills, steps and rubble, and ones that had the complexities of the second sort, but with added props like cubes that needed to be moved around.
它們漫游其中的環(huán)境也很簡(jiǎn)單，分為三種：第一種很平坦，第二種布滿小山丘、臺(tái)階和瓦礫；第三種和第二種一樣復(fù)雜，但還增加了像立方體這樣需要被移動(dòng)的道具。
To begin with, the unimals were given a variety of randomly assigned shapes, but with identical software running each of them.
首先，研究人員給unimal提供了各種隨機(jī)分配的形狀，但每個(gè)形狀都有相同的軟件運(yùn)行。
That software was a piece of artificial intelligence called a deep evolutionary reinforcement learning algorithm, or derl.
該軟件是一款人工智能軟件，叫做深度進(jìn)化強(qiáng)化學(xué)習(xí)算法，簡(jiǎn)稱derl。
Newly created unimals started in a virtual boot camp, in which the derl learned enough about the world to face the challenges to come.
新創(chuàng)建的unimal始于一個(gè)虛擬的新兵訓(xùn)練營(yíng)，derl可以對(duì)世界有足夠多的了解以應(yīng)對(duì)即將到來(lái)的挑戰(zhàn)。
They were then entered into tournaments.
然后開(kāi)始進(jìn)行錦標(biāo)賽。
In groups of four, Dr Gupta put them through tests of agility, stability and ability to manipulate objects.
古普塔博士將他們分成四人一組，讓他們接受敏捷性、穩(wěn)定性和操縱物體能力的測(cè)試。
Each group's winner was allowed to "breed" by spawning a daughter with one mutation (an extra limb for stability, perhaps, or extra rotation in a joint, for flexibility).
每組的優(yōu)勝者都被允許“繁殖”，產(chǎn)生一個(gè)帶有突變的后代（可能是為了穩(wěn)定，或者是為了關(guān)節(jié)的額外旋轉(zhuǎn)，以提高靈活性）。
This daughter was substituted for the oldest unimal in the pool, assigned to a new group of four, and the process repeated.
這個(gè)后代取代了其中年齡最大的一個(gè)，被分配到一個(gè)新的四人組，這個(gè)過(guò)程不斷重復(fù)。
Unimals were withdrawn from the fray after ten generations of evolution, and Dr Gupta reckons about 4,000 varieties of them underwent training.
經(jīng)過(guò)十代的進(jìn)化后結(jié)束比賽，古普塔博士估計(jì)它們中大約有4000個(gè)變種接受了訓(xùn)練。
The team were surprised by the diversity of shapes that evolved.
研究小組對(duì)進(jìn)化過(guò)程中形狀的多樣性感到驚訝。
Some had arms as well as legs.
有些既有胳膊又有腿。
Others had only legs.
其他的只有腿。
There were bipeds, tripeds and quadrupeds.
有兩足、三足和四足動(dòng)物。
Some moved like lizards.
有些像蜥蜴一樣移動(dòng)。
Others resembled an octopus walking on land.
有些像在陸地上行走的章魚(yú)。
Crucially, though, the researchers found that the most successful unimals learned tasks in half the time that their oldest ancestors had taken, and that those which evolved in the toughest arenas were the most successful of all.
然而，至關(guān)重要的是，研究人員發(fā)現(xiàn)，最成功的unimal學(xué)習(xí)任務(wù)的時(shí)間是它們最古老的祖先所用時(shí)間的一半，而那些在最艱苦的環(huán)境中進(jìn)化出來(lái)的unimal是最成功的。
In this evolution of unimals' morphology to promote the ability to learn, Dr Gupta sees a version of something called the Baldwin effect.
古普塔博士看到了鮑德溫效應(yīng)在競(jìng)賽中的體現(xiàn)，在unimal的進(jìn)化過(guò)程中促進(jìn)了學(xué)習(xí)能力的提高。
In 1896 James Baldwin, an American psychologist, argued that minds evolve to make optimal use of the morphologies of the bodies they find themselves in.
1896年，美國(guó)心理學(xué)家詹姆斯·鮑德溫提出，人的進(jìn)化是為了最大限度地利用他們所處的身體形態(tài)。
What Dr Gupta has shown, though in software rather than in the real, biological world, is that the obverse can also be true—changes in body morphology can optimise the way minds (or, at least, derls) work.
古普塔博士已經(jīng)證明，盡管是在軟件中而不是在真實(shí)的生物世界中，相反的情況也可能會(huì)發(fā)生的——身體形態(tài)的改變可以優(yōu)化大腦（或者至少是derl）的工作方式。
Even though he held the software constant from generation to generation, it became more efficient at learning as the unimals' bodies evolved.
盡管他一代又一代地保持軟件不變，但隨著unimal的進(jìn)化，它在學(xué)習(xí)方面也更加高效。
Whether that discovery can be turned to account in the way robots are developed remains to be seen.
這一發(fā)現(xiàn)是否可以用于機(jī)器人的開(kāi)發(fā)，還有待考究。
But it is certainly, in the jargon beloved of some businessfolk, an out-of-the-box idea.
但用一些商人喜愛(ài)的行話來(lái)說(shuō)，這肯定是一個(gè)創(chuàng)新的想法。
