双语|农业机器人将如何颠覆传统农业？

(本文选自《经济学人》20200208期)

背景介绍：

随着人工智能技术的不断发展，越来越多的行业利用人工智能技术实现跨越性地发展，而农业也不例外。农业与人工智能的相结合正在解放农民的双手，并使得精准农业的愿景得以实现。让我们一起来看看英国的这家公司是如何通过人工智能技术来颠覆传统农业的。

Using artificial intelligence, agricultural robots are on the rise

农业机器人正在借助于人工智能技术而崛起

Slowly but surely, a spidery machine about the size of a golf cart swings an electrode over a tray of plants. Every few seconds there is a small puff of smoke as a weed keels over, having been zapped with a high voltage.

一台高尔夫球车一般大小的蜘蛛机器人虽然缓慢但却很稳当地在一撮植物上摆弄着电极。每隔几秒，杂草就会被高压电击中，并冒出一小股烟。

The device doing the zapping is a prototype weeding robot developed by the Small Robot Company, a new firm operating out of an old munitions depot near Salisbury, in south-west Britain.

该除草设备是由小型机器人公司（一家开在英国西南部索尔兹伯里附近的一个废旧军火库里的公司）开发的原型除草机器人。

Weed control is essential for improving crop yields, but it is getting increasingly difficult. Some weeds are becoming resistant to herbicides, which face stricter regulation and in some cases are being banned. On top of that, many consumers want organic produce.

想要提高农作物的产量，除掉杂草是关键，但这正在变得越来越困难。一些杂草正在对除草剂产生抗药性，而除草剂正面临着更严格的监管，在某些情况下往往还被禁止使用。除此之外，许多消费者也想吃有机农产品。

And labour shortages mean that repeatedly tilling the soil to disrupt weed growth using a mechanical hoe towed behind a tractor is costly, time consuming and not always practical.

而劳动力的短缺意味着，利用拖拉机后面拖着的机械锄头反复耕地的方式来除草不仅成本高昂、费时，而且效果也不尽如人意。

Weeding is a chore that most farmers would happily hand to robots. But for a robot to do the job properly it must be able to distinguish a weed from what is being cultivated. That is becoming easier with advances in computer vision. Artificial-intelligence (AI) algorithms are getting better at classifying images.

除草是一件苦差事，大多数农民都乐意将其交给机器人干。但想让机器人能够完美地完成这项工作，首先它必须得分辨出哪些是杂草、哪些是作物。随着机器视觉的发展，这一点越来越容易实现了。人工智能算法在图像分类方面做得越来越好了。

Some phone apps can now identify a plant from just a photo. Robots equipped with cameras will not only weed but automate other farming roles. Agribots, driverless tractors and other types of farm automation form an industry that is expected to grow at around 23% a year and to be worth more than $20bn by 2025, according to MarketsandMarkets, an American research firm.

如今，一些手机应用程序甚至可以识别照片中的植物。配备有摄像头的机器人不仅能够除草，还能使其他农活实现自动化。据美国研究公司 MarketsandMarkets 的数据显示，农业机器人、无人驾驶拖拉机和其他类型的农业自动化正在形成一个以每年23%的速度增长的产业，预计到2025年，该产业的价值将超过200亿美元。

Having spotted a weed, there are several ways to try to kill it. The Small Robot Company’s weeding agribot, called Dick, electrocutes them. The robot’s wheels work like an electrode to make contact with the ground while another electrode is moved to touch the plant.

在识别杂草后，有几种方式可以将其除掉。小型机器人公司开发的这款除草机器人名叫迪克，它使用的是电击的方式。机器人的轮子作为一根电极与地面接触，同时另一根电极通过移动与杂草接触。

This makes a circuit through the plant and creates heat, effectively boiling the plant’s cells and killing it from stem to root instantly. It can take several thousand volts, although this is adjusted according to the type of weed. The remnants of the plant can then be left to naturally decompose into the soil.

这便通过杂草形成了一个回路，同时产生热量，使植物细胞瞬间气化，从而从茎至根将杂草除去。这或许得高达几千伏的电压，当然也根据杂草类型的不同而有所差异。随后，杂草的残余物将在土壤中自然分解。

Dick will work with two other agribots, Tom and Harry, all of them electrically powered. The idea, says Ben Scott-Robinson, one of the company’s co-founders, is that Tom will routinely scan fields to a level of detail of just a few centimetres.

迪克将与另外两个电动农业机器人（汤姆和哈里）一起工作。该公司联合创始人之一的本·斯科特-罗宾逊表示，他们的想法是，汤姆负责进行常规扫描，其扫描精度甚至可以达到厘米级。

Using its cameras and other sensors, the spindly four-wheel-drive machine can cover about 20 hectares a day, mapping the health of every plant, along with soil conditions. When weeds appear, Dick will be dispatched to zap them. Harry, the third agribot, is under development for tasks such as seeding and applying precise levels of fertiliser to each plant, a process known as microdosing.

这台细长的四轮驱动机器人每天扫描土地的面积可达20公顷左右，它将绘制出每株植物的健康状况以及土壤状况。当发现杂草时，迪克将被派去将其消灭。第三台名为哈里的农业机器人目前仍在研发中，其任务包括播种以及对每株植物进行精确施肥（这一过程被称为微量施肥）。

Tom will go into commercial production in August. Early versions are already in use on some farms in Britain, including the Leckford Estate, which is run by a large supermarket chain. Dick will begin trials later this year. Tom has already learnt how to spot broadleaved weeds and is being trained in the more difficult art of identifying grass-type weeds, which look similar to cereals.

汤姆将于今年8月投入商业生产。其早期版本已经被应用于英国的一些农场了，其中包括由大型连锁超市经营的莱克福德庄园。迪克将于今年晚些时候开始进行试验。汤姆已经学会了如何识别阔叶杂草，并且正在接受难度更大的杂草类型识别技术的训练。

Unlike many vegetables, cereals are grown more densely in rows, which makes the mechanical hoeing of weeds a tricky operation. Armed with Tom’s algorithms, Dick should be able to zap grassy weeds in cereals accurately.

谷类作物不同于大多数蔬菜，其种植密度更大，这也使得机械除草变得十分困难。但借助于汤姆的算法，迪克应该就能准确地除掉谷物中的杂草了。

声明：本媒体部分图片、文章来源于网络，不代表本人观点，版权归原作者所有，如有侵权，请联系本站删除。