Self-Assembling Robots and the Potential of Artificial Evolution | Emma Hart | TED

翻译人员: Alvin Lee 校对人员: Yip Yan Yeung

Imagine a scientist

假设一位科学家

who wants to send a robot to explore in a faraway place,

想送一台机器人去探索很远的星球，

a place whose geography might be completely unknown

那里的地理环境可能完全未知，

and perhaps inhospitable.

可能不宜居。

Now imagine that instead of first designing that robot

想象一下，如果我们 不是先设计好这个机器人，

把它发射出去，希望它能适应那里，

and sending it off in the hope that it might be suitable,

instead, she sends a robot-producing technology

而是发射一台生产机器人的机器人，

that figures out what kind of robot is needed once it arrives,

抵达后就能搞清楚 那里需要什么样的机器人，

builds it and then enables it to continue to evolve

把它建造出来， 并让它能够持续进化，

以适应新的环境。

to adapt to its new surroundings.

It’s exactly what my collaborators and I are working on:

这正是我和我的同伴们研究的课题：

a radical new technology which enables robots to be created,

一项全新的技术， 能让机器人被创造，

reproduce and evolve over long periods of time,

被复制并随时间进化，

a technology where robot design and fabrication becomes a task

这项技术能让机器人的设计和制造

for machines rather than humans.

由机器而不是人类来完成。

Robots are already all around us, in factories, in hospitals, in our home.

机器人在工厂、医院、 家庭已经随处可见。

But from an engineer's perspective,

但从工程师的角度来看，

designing a shelf-stacking robot or a Roomba to clean our home

设计自动补货机器人或者扫地机器人

is relatively straightforward.

相当直接了当。

我们非常清楚应该怎么做，

We know exactly what they need to do,

and we can imagine the kind of situations they might find themselves in.

我们可以设想机器人 可能会面临的情况。

So we design with this in mind.

在设计时记住这一点就行了。

但是，如果我们对 机器人将要工作的地点

But what if we want to send that robot to operate

in a place that we have little or even no knowledge about?

所知甚少甚至一无所知呢？

For example, cleaning up legacy waste inside a nuclear reactor

比如，要清除 核反应堆中的残留废料，

where it's unsafe to send humans,

送人进去肯定是不安全的，

mining for minerals deep in a trench at the bottom of the ocean,

比如要在海洋底部的 海沟中开采矿石，

or exploring a faraway asteroid.

或者探索一颗遥远的小行星。

How frustrating would it be if the human-designed robot,

如果人类设计的机器人

花了好几年才抵达小行星，

that had taken years to get to the asteroid

suddenly found it needed to drill a hole

突然发现需要钻洞取样，

to collect a sample or clamber up a cliff

或者需要攀爬上一个悬崖，

but it didn't have the right tools

但却没有合适的工具

or the right means of locomotion to do so?

或者合适的移动方式来实现， 该多么令人郁闷。

If instead we had a technology

如果取而代之的是， 我们拥有一种技术，

that enabled the robots to be designed and optimized in situ,

能让机器人根据它们 将要生活和工作的环境

in the environment in which they need to live and work,

被设计和优化，

then we could potentially save years of wasted effort

我们可能能节省数年被浪费的努力，

and produce robots that are uniquely adapted

生产出独一无二的、

to the environments that they find themselves in.

能适应它们所处环境的机器人。

So to realize this technology, we've been turning to nature for help.

为了实现这项技术， 我们转向大自然寻求帮助。

All around us,

环顾我们四周，

we see examples of biological species

我们能看到各种生物，

that have evolved smart adaptations

进化出智能的适应性，

that enable them to thrive in a given environment.

能让他们在特定的环境中蓬勃发展。

比如，在古巴的雨林中，

For example, in the Cuban rainforest,

we find vines that have evolved leaves

我们发现一种藤蔓植物，它的叶子

that are shaped like human-designed satellite dishes.

进化成了人类设计的 碟形卫星天线的形状。

These leaves direct bats to their flowers

这些叶子能放大蝙蝠发出的信号，

by amplifying the signals that the bats send out,

将它们引向自己的花，

therefore, improving pollination.

从而促进授粉。

What if we could create an artificial version of evolution

如果我们能创造出人工版本的进化，

that would enable robots to evolve in a similar manner

让机器人以类似生物有机体的方式

as biological organisms?

来进化呢？

我所说的并不是仿生学，

I'm not talking about biomimicry,

a technology which simply copies what's observed in nature.

仿生学只是简单模仿 我们在自然界观察到的事物。

What we're hoping to harness is the creativity of evolution,

我们希望利用的是进化的创造力，

去发现我们在地球上 没看到过的设计，

to discover designs that are not observed here on Earth,

the human engineer might not have thought of

人类工程师也许没有想到过，

甚至没有能力想象出来这样的设计。

or even be capable of conceiving.

In theory,

理论上来说，

这一革命性的设计技术能够在遥远的星球

this evolutionary design technology could operate completely autonomously

完全自主运行。

in a faraway place.

But equally it could be guided by humans.

但同样的，它也能够由人类指导。

Just as we breed plants for qualities such as drought resistance or taste,

就像我们培育植物， 以期具备抗旱能力或者改良味道，

the human robot breeder could guide artificial evolution to producing robots

人类中的机器人培育者 也可以指导人工进化，

来制造具有特定特性的机器人。

with specific qualities.

For example,

比如，

the ability to squeeze through a narrow gap

能挤过狭窄的缝隙，

or perhaps operate at low energy.

或者在低能量情况下运行。

This idea of artificial evolution imitating biological evolution

这种通过模仿生物进化

using a computer program

来打造人造进化的想法—— 通过计算机程序

to breed better and better solutions to problems over time

一次次优化解决方案——

isn't actually new.

并不是新的。

In fact, artificial evolution,

实际上，人造进化，

algorithms operating inside a computer,

计算机内部运行的算法，

have been used to design everything from tables to turbine blades.

已经可以用来设计一切了， 无论是桌子还是涡轮叶片。

Back in 2006,

早在 2006 年，

美国宇航局将一颗 人造卫星送入太空，

NASA even sent a satellite into space with a communication antenna

卫星上搭载的通信天线， 就是由人工进化设计的。

that had been designed by artificial evolution.

But evolving robots is actually much harder

但让机器人进化

than evolving passive objects such as tables,

比让被动的物体—— 比如桌子——进化要难得多，

because robots need brains as well as bodies

因为机器人不光需要身体， 还需要大脑，

in order to make sense of the information in the world around them

用来处理它们周围的信息

and translate that into appropriate behaviors.

并采取相应适当的行为。

So how do we do it?

那我们该如何做呢？

Surprisingly, evolution only needs three ingredients:

意外的是，进化只需要 3 个条件：

a population of individuals which exhibit some physical variations;

一定数量的族群， 能展现出个体上的差异；

a method of reproduction

一种繁殖方式，

in which offspring inherit some traits from their parents

后代能继承父母的某些特点，

and occasionally acquire new ones via mutation;

偶尔通过变异能产生新的特点；

and finally, a means of natural selection.

最后，一种自然选择的方法。

So we can replicate these three ingredients to evolve robots

因此我们能用软硬件结合的方式

using a mixture of hardware and software.

来复制这 3 个条件， 使机器人进化。

The first task is to design a digital version of DNA.

首先需要设计 一种数字版本的 DNA。

这是一份数字蓝图， 记录了机器人的大脑、身体、

That is a digital blueprint that describes the robot's brain, its body,

its sensory mechanisms and its means of locomotion.

感觉器官和运动方式。

Using a randomly generated set of these blueprints,

随机生成几组这样的蓝图，

we can create an initial population of 10 or more robots

我们就能创造出最初的 10 个或者更多机器人，

to kick-start this evolutionary process.

来启动这个进化的过程。

We've designed a technology that can take the digital blueprint

我们已经设计出一种技术， 能运用这些数字蓝图

and turn it into a physical robot without any need for human assistance.

建造出机器人实体， 中间完全不需要人类的协助。

For example, it uses a 3D printer to print the skeleton of the robot

比如，它可以 3D 打印出机器人的骨骼，

and then an automated assembly arm like you might find in a factory

然后一条自动组装机械臂， 就像你在工厂里见过的那种，

to add any electronics and moving parts,

给它装上电路和可以动的零件，

including a small computer that acts as a brain.

包括一台小型计算机充当它的大脑。

And to enable this brain to adapt to the new body of the robot,

然后让这个大脑适应机器人的新身体，

we send every robot produced to an equivalent of a kindergarten,

我们将生产出来的每一台机器人 送到一个相当于幼儿园的地方，

a place where the newborn robot can refine its motor skills

新生的机器人可以在那里 锻炼它们的运动技能，

almost like a small child would.

就像小孩子一样。

To mimic natural selection,

为了模拟自然选择，

we score these robots on the ability to conduct a task.

我们会给机器人 处理任务的能力打分。

And then we use these scores

根据这些分数

to selectively decide which robots get to reproduce.

来决定哪些机器人需要回炉重造。

The reproduction mechanism

这一重造机制

mixes the digital DNA of the chosen parent robots

将被选中的父母机器人的 数字 DNA 混合，

to create a new blueprint for a child robot

为机器人孩子重新绘制一张蓝图，

这些机器人孩子 继承了父母的一些特征，

that inherits some of the characteristics from its parents

but occasionally also exhibits some new ones.

但偶尔也会展现出新的特征。

And by repeating the cycle of selection and reproduction over and over again,

通过不断重复这个 从选择到重造的循环，

we hope that we can breed successive generations of robots

我们希望能够培育出一代代机器人，

where, just like is often observed in biological evolution,

就像我们在生物进化中 经常观察到的，

each generation gets better than the last,

每一代都比前一代要好，

with the robots gradually optimizing their form and their behavior

这样一来，机器人能逐渐优化 它们的形态和行为，

to the task and the environment that they find themselves in.

以更好地适应面临的任务 和所处的环境。

Now, although this can all take place

尽管这一切

in a time frame that's much faster than biological evolution,

比起生物进化而言要快得多，

which sometimes takes thousands of years,

后者有时需要数千年之久，

it's still relatively slow in terms of the time frames we might expect

但相对而言，这一过程 仍然比我们期待的要慢，

in our modern world

在我们这个时代，

我们希望设计和制造 某种产品的速度非常快。

to design and produce an artifact.

It's mainly due to the 3D printing process,

主要是因为 3D 打印的过程较慢，

which can take more than four hours per robot,

每个机器人需要 超过 4 小时的时间，

depending on the complexity and the shape of the robot.

取决于机器人的复杂程度和形状。

But we can give our artificial evolutionary process a helping hand

但我们可以给人工进化提提速，

减少我们实际需要制造的 实体机器人数量。

to reduce the number of physical robots that we actually need to make.

We create a digital copy of every robot produced

我们可以在计算机的模拟程序中

inside a simulation in a computer,

制作每一台机器人的数字版本

然后让这虚拟的机器种族进化。

and we allow this virtual population of robots to evolve.

Now it's quite likely that the simulation isn't a very accurate representation

当然，这种模拟并不能很精准地反映

of the real world.

真实世界。

But it has an advantage that it enables models of robots to be created

但它的优势在于能在几秒钟 内将机器人制造出来并进行测试，

and tested in seconds rather than hours.

而不需要花费数小时。

因此运用这种模拟技术，

So using the simulator technology,

we can quickly explore the potential of a wide range of robot types

我们能很快速地 探索多种机器人的潜力，

of different shapes and sizes, of different sensory configurations,

它们形状不同，大小各异， 搭配的传感器也不一样，

and quickly get a rough estimate of how useful each robot may be

我们在造出机器人实体之前， 就能大概估算出

before we physically make it.

它的用处能有多大。

And we predict that by allowing a novel form of breeding

我们预测，通过这种新的培育方法，

in which a physical robot can breed with one of its virtual cousins,

实体的机器人能够同它的 虚拟表兄弟一起被培育，

然后在模拟中被发现的有用的特征

then the useful traits that have been discovered in simulation

will quickly spread into the physical robot population,

将会迅速扩散到实体机器人族群中，

where they can be further refined in situ.

它们可以在现场得到进一步优化。

It might sound like science fiction,

这听起来也许像科幻小说，

but actually there's a serious point.

但实际上是一个很严肃的问题。

While we expect the technology that I've just described

当我们期待刚刚我描述的技术

to be useful in designing robots,

能在机器人设计中有用的时候，

for example, to work in situations where it's unsafe to send humans

比如，让机器人到对人类而言 很危险的地方工作，

or to help us pursue our scientific quest for exoplanetary exploration,

或者帮助我们实现科学目标， 用于系外行星的探索，

there are some more pragmatic reasons

有更多现实的原因，

让我们应该考虑人工进化。

why we should consider artificial evolution.

As climate change gathers pace,

随着气候变化的加速，

it is clear that we need a radical rethink

我们需要重新思考

to our approach to robotic design here on Earth

如何在地球上设计机器人，

才能减少生态足迹。

in order to reduce that ecological footprint.

For example,

比如，

creating new designs of robot built from sustainable materials

设计新的由可持续 材料制造的机器人，

能以较低能耗运行，

that operate at low energy,

that are repairable and recyclable.

能够维修和可回收。

It's quite likely that this new generation of robots

很显然，这种新一代的机器人

won't look anything like the robots that we see around us today,

跟我们今天见到的机器人完全不同，

but that's exactly why artificial evolution might help.

但这恰恰是人工进化 也许能起到作用的原因。

打破我们自己对于 工程科学理解的局限，

Discovering novel designs by processes that are unfettered by the constraints

that our own understanding of engineering science

减少这种局限 对设计过程造成的约束，

imposes on the design process.

我们就能发现新颖的机器人设计。

Thank you.

谢谢大家。

(Applause)

（掌声）