Pick­ing ro­bots in Eng­land

A lot of money is be­ing in­vested in farm au­to­ma­tion re­search in Eng­land.

The Orchardist - - Technology - By Dan Bloomer, LandWISE

An added im­pe­tus ap­pears to be fear sur­round­ing Brexit. As in so many places, im­mi­grant labour per­forms many ba­sic jobs in hospi­tal­ity and of course, agri­cul­ture and hor­ti­cul­ture. They fear a post-Brexit re­duc­tion in the avail­abil­ity of East Euro­pean labour.

In the next few years, I be­lieve au­to­ma­tion tech­nolo­gies will pro­vide most ben­e­fit through mak­ing work­ers more ef­fi­cient, not re­plac­ing them en­tirely. Some jobs are easy for a ro­bot and hard for a per­son – think build­ing cars on assem­bly line. Peo­ple have been largely re­placed by ma­chines that do the same very pre­dictable job over and over very ac­cu­rately.

Other jobs are easy for a per­son but very hard for a ro­bot – think pick­ing ap­ples! This highly vari­able task is easy for a per­son but ex­tremely tricky for a ro­bot. “See­ing” is hard enough, get­ting a pick­ing arm to the fruit, hold­ing and dis­con­nect­ing it from the tree is quite an­other. Even get­ting to the tree is a challenge for a ro­bot. One of the world’s largest veg­etable crops is broc­coli, al­most en­tirely man­u­ally har­vested. Tom Duck­ett at the Univer­sity of Lincoln in Eng­land leads a group de­vel­op­ing a ma­chine har­vester. He says their sys­tem is as good as hu­man pick­ers at de­tect­ing broc­coli heads of the right size. With iden­ti­fi­ca­tion in hand, devel­op­ment is now on me­chan­i­cal cut­ting and col­lect­ing. Then you need the ro­bot that car­ries the pick­ing ro­bot around the field.

Ma­chine de­tec­tion of broc­coli (and prac­ti­cally any­thing else) is best if the ro­bot can pick at night when con­trolled light­ing makes image anal­y­sis much eas­ier. This may be a pos­i­tive fea­ture: ro­bots read­ily work in the cool of the night with sig­nif­i­cant crop qual­ity ben­e­fits though re­duced field heat. It also sep­a­rates ma­chines from peo­ple, one of the health and safety is­sues of con­cern with driver­less ma­chines.

The hu­man hand can eas­ily pick up, move and place ob­jects, but for a ro­bot, grip­ping ef­fec­tively is a challenge. An­dre Rosendo at the Univer­sity of Cam­bridge says some­times it is not the best op­tion to try to build some­thing that is “like a hu­man” – but in­stead de­sign the best ma­chine for the job.

An early pro­to­type us­ing soft ro­bot­ics for let­tuce har­vest­ing aimed to mimic the ac­tion of a hu­man hand grasp­ing the let­tuce head, tilt­ing it, and then cut­ting the stalk. How­ever a real challenge has been the in­abil­ity of the “hand” to grasp the let­tuce with enough strength to sub­se­quently lift it.

US re­searchers by­passed tra­di­tional “hand” de­signs, and cre­ated a ver­sa­tile grip­per us­ing every­day ground cof­fee and a la­tex party bal­loon. The cof­fee-filled bal­loon presses down and de­forms around the ob­ject, then a vac­uum sucks the air out of the bal­loon, so­lid­i­fy­ing its grip. When the vac­uum is re­leased, the bal­loon be­comes soft again, and the grip­per lets go.

Harper Adams Univer­sity in Shropshire is de­vel­op­ing a ro­botic straw­berry har­vester. They have not de­signed a sim­ple picker re­place­ment. Devel­op­ment has re­quired in­te­grat­ing ge­netic devel­op­ment for va­ri­eties with long stalks, a grow­ing sys­tem that has plants off the ground, and the ro­botic tech­nolo­gies to iden­tify, lo­cate and as­sess the ripeness of in­di­vid­ual straw­ber­ries and then pick them touch­ing only the pe­dun­cle (stalk).

Orchard tasks like grass mow­ing are rel­a­tively easy to au­to­mate. And var­i­ous groups have done so. Fruit pick­ing has to be one of the big­gest chal­lenges of all. Ro­bot­ics does have a huge po­ten­tial but we are not there yet. It’s about re­think­ing the whole sys­tem not just re­plac­ing the hu­man picker. But we can use sim­ple me­chan­i­cal aides to as­sist work­ers. Prun­ing and pick­ing plat­forms have been avail­able for a long time. As well as get­ting the picker to the spot, they carry the load and avoid ex­tra han­dling. And plat­forms can eas­ily be made much smarter: ex­ist­ing tech­nol­ogy could guide them along rows and we could fit sen­sors to pre-grade and sep­a­rate fruit im­me­di­ately.

Si­mon Pearson at the Na­tional Cen­tre for Food Man­u­fac­tur­ing says, “It’s a Franken­stein thing, this agrobotics. There are all sorts of great bits avail­able but you have to seek them out and stitch them to­gether your­self to make the crea­ture you want.”

From left: While peo­ple can nav­i­gate an orchard, find and pick ap­ples, it is a fiendishly dif­fi­cult task for a ro­bot. “Tomi” by Harper Adams Univer­sity is one of the ro­botic grass mow­ers de­vel­oped. Typ­i­cally stan­dard mow­ers are given lo­ca­tion and contr

Newspapers in English

Newspapers from New Zealand

© PressReader. All rights reserved.