THEME | ROBOTS

Robots for the future

Environmental consciousness has been the big motivator in the Danish development of innovative robots for agriculture

By Nadia Louise Kristensen

Illustration Lars Chrois

In the future, it won’t be human beings who will bend their backs again and again to pull out weeds from between the crops on organic farms and outdoor nurseries. In most places robots will have replaced the strenuous work that does not require the special human quality of empathy. And the robots will run on renewable energy. More conventional farms will follow in the footsteps of organic farms and replace sprays with weeding robots, because it is better for the environment and the crops. This is how the future can look. And it is worth it because consumers in the western world continue to demand products which have been produced under sustainable conditions. The technology is under development, and Danish research scientists and companies are among the leaders in this field.

“Denmark has comprehensive agricultural experience, and research is well advanced in this area. We concentrated on organic farming at an early stage and have focused on reducing the environmental impact of agriculture, both to minimize the farmer’s costs and protect the environment,” says Niels Jul Jacobsen, a project manager at RoboCluster, an innovation network and one of the places in Denmark where work is constantly under way to develop robots – including robots for agriculture.

Sensing like a human being

Denmark has a long tradition for sustainability – in the energy sector, the development and sales of wind turbines is perhaps one of the most successful export stories in Danish history.

Since the late 1990s, a number of Danish research scientists have been working on making agricultural robots that can assist or replace human beings in weeding, sowing or spraying tasks.

“It is the environmental perspective which has provided the momentum for many of the research projects. At the end of the 1990s, researchers started getting these ideas about robots. They could see that we had both the computer technology and the camera technology – the trick was to combine them in something that could be used in agriculture,” says Michael Nørremark, a researcher at the Department of Agricultural Engineering at the University of Aarhus.

Michael Nørremark has been part of developing an agricultural robot intended for use out on the fields, namely the Hortibot, which was judged Best Invention Of The Year in 2007 in America’s Times Magazine. The 1.2 metre high Hortibot weighs 200 kilos and is a driverless weed fighter.

It has been and still is a challenging process to develop Hortibot and other types of agricultural robots to weed, sow or spray in the fields, which are filled with countless obstacles of every conceivable kind.

“The machine needs to be able to run by itself for many hours without human interference. The big challenge is to get the robots to sense what we humans do when we are out in the fields. We look, listen and sense, all at the same time,” says Michael Nørremark.

Despite their similarities, plants can also have many different forms and colours. The research scientists have thus put a lot of work into developing camera technology and software so advanced that the robots can match human vision and intelligence.

“The factory robots we know today typically work in highly structured surroundings. They are often enclosed in a limited area and programmed to stop if they encounter an unknown situation. A farmer’s field is different. The robots will often encounter things that they have no knowledge of in advance. So they have to be able to handle the unknown. It is a major challenge,” says Niels Jul Jacobsen.

Several research teams have now developed camera technology and software which to some extent can distinguish between different types of plants and weeds. How this has been done at RoboCluster is still a secret because the method is in the process of being patented.

Aiming for internationalisation

One of the things that the RoboCluster innovation network does is to involve all the parties needed to develop and commercialise the robots. The network builds a bridge between suppliers, users and research and educational institutions specialising in robot and automation technology.

“We create synergy between research, industry and society and pave the way for innovation through development projects and professional networks. Our experience shows that it is important to include users in the development process, while companies have the commercial knowledge and the researchers provide the latest scientific knowledge,” says Lasse Mogensen, director of RoboCluster.

The network is very focused on international perspectives and hopes to become a natural international collaboration partner for the development of robot technology.

“We hope that robot developers in Japan, France and the USA will contact us regarding research, development or collaboration on robot technology. We are working on that objective by developing and expanding our international collaboration through international conferences and international projects, where we bring the competences and opportunities of the RoboCluster network into the arena,” says Lasse Mogensen.

Interaction between man and robot

RoboCluster is also working on different kinds of agricultural robot. The network has for example developed Agrobot, which is based on the same mobile platform as Hortibot, but has different tools. Two versions have been studied – one which like Hortibot can remove weeds mechanically, and another that kills weeds with sprays. Keld Bertelsen, an industrial designer at RoboCluster, has been working on Agrobot and has created the design concept for the fully automatic field robot Roboss, which is based on the same technology as Agrobot. In the development process, Keld Bertelsen considers both the user’s needs and the function of the robot.

“It is an interplay between what is going on in the field and the robot and the human being. I need to find out about how farmers can use this robot. One of the greatest challenges in the design process is to understand their way of thinking and their way of thinking about robots,” says Keld Bertelsen.

Having interviewed a number of farmers, Keld Bertelsen realised that organic farmers and conventional farmers might differ in the type of robot they need.

“The organic and biodynamic farmers have chosen the farming sector because they like being in contact with plants and nature. So it seemed wrong to me to base my concept on robots exclusively being controlled from a computer in the farm office. Instead I developed a concept where the farmer carries out the task in collaboration with the field robot – where the robot becomes an agricultural worker that is accompanied out into the field on the first occasion and shown where to work. In that way the farmer also gets out to see how the plants are faring,” says Keld Bertelsen.

Agrobot and Roboss can find their way around the field by themselves, identify weed species and give unwanted plants carefully dosed sprays without contaminating the crop that is being cultivated. According to Niels Jul Jacobsen, the usage of sprays can be reduced by 90 per cent with the new method.

One of the advantages of the robots is that they can be exploited better than a large tractor. For example, they are better at manoeuvring in corners and on bumpy ground. And they do not churn up the soil like big machines do.

“When you are driving a tractor, the soil is displaced along the line in which you are driving. And the bigger the tractor is, the broader the tyres and the greater the area you cannot cultivate,” says Niels Jul Jacobsen.

Top: Hortibot was judged Best Invention of the Year in 2007 by Time Magazine
Bottom: Agrobot is being developed by RoboCluster in Denmark.

Ahead of its time

In a way, the development of robots for agriculture is slightly ahead of its time. Although there will be buyers among organic farms it is not certain, when the costs are reckoned up, that it is worth it for conventional farms to replace chemistry with remote-controlled weed fighters – yet.

“There is no market for it now, because sprays are so cheap. But we know that environmental requirements will come which will create a market in the future,” says Niels Jul Jacobsen.

The target group for field robots is thus initially organic farms and outdoor nurseries, where the crop has a higher value per hectare than in conventional agriculture.

“New technology is always like this – expensive to produce at the beginning. There is no hope of Hortibot outcompeting large spraying machines on price, but we have created an alternative,” says Michael Nørremark.

Hortibot has reached a stage where it can handle simple tasks such as weeding rows, and where the crops are visible. But Michael Nørremark and the rest of the research team are continuing their work to develop the machine’s artificial intelligence.

“In the long term, the aim is that the robot can find the way to the field by itself and make a plan for sowing a field which has just been ploughed, but there is some way to go yet,” says Michael Nørremark.

Hortibot is equipped at present with an internal combustion engine, but the Hortibot of the future will be more environmentally friendly.

“In the future vehicles will not run on fossil fuels but on electricity, renewable energy or biofuel,” concludes Michael Nørremark.

Live-reportage on fd.denmark.dk

Robots to weed

No more is needed than a small Danish company and a good idea to create technology which can lead to growth and a better agricultural environment

By Nadia Louise Kristensen

Top right: The mechanical pincer robot from FP Engineering is ready to go on sale in 2009
Bottom left:The non-toxic flame-weeder from FP Engineering has been ten years in the making. Hopefully it will be ready for sale in 2010.

Switch on, switch off, switch on, switch off. Small plasma gas flames are switching on and off at a furious pace. The flames measure just 10 millimetres in diameter and are only 100 millimetres long. They switch off when they pass an onion and switch on again when they encounter a weed. The gas burner is part of a brand new agricultural robot which can remove weeds. The method is non-toxic and has taken Danish company FP Engineering nearly ten years to develop. Thermal precision weeding is the official term for this method. The company just lacks the last adjustments to make the burner stable, but in about a year it should be ready for sale.

One of Denmark’s leading outdoor nurseries has tested the robot on its onion fields, and has great expectations of what it can achieve.

“We can use robots to rationalise our working procedures in production, so that fewer man hours are needed and we will be less dependent on labour. We can produce the bulk of our production organically, and at a lower price,” says Axel Månsson, who owns the outdoor nursery Månssons.

The target group is primarily countries where labour is so expensive that it is worth it to replace human beings with robots.

The two biggest challenges have been to get the burner to switch on and off at high speed, and to get the robot to distinguish between weeds and onions.

“We cannot get the robot to recognise all types of weeds. That is virtually impossible. But we can get the robot to recognise the special features of onions. There was no camera on the market that we could use, so we had to develop a special plant camera ourselves,” says Frank Poulsen, who is an entrepreneur and the owner of FP Engineering.

The challenge is that you cannot control sunlight. Colours change depending on whether the sun is rising, high in the sky, or setting. The solution was a camera with a special lens which will only look for chlorophyll and not be distracted by soil or stones or other things such as straw or dead plant remains which often lie on the fields.

Robots replace pesticides

The obvious target group is organic farms, but Frank Poulsen believes that robots which can remove weeds mechanically
or with a burner can outcompete sprays.

“When you are spraying, you are in principle just applying a substance that the utility plant can tolerate and which destroys weeds. But it is a modified truth that the utility plant can tolerate it. The situation is similar to giving a cancer patient chemotherapy. It kills the cancer, but it also weakens the patient significantly.

It is a question of price. If it is worth it for growers to use robots instead of sprays, they will do it. They cannot afford to lag behind,” says Frank Poulsen.

He has also developed an agricultural robot which can remove weeds with a pincer that moves in and out of the rows. It uses the same method as the flame robot, to localise what is a weed and what is a crop plant. But instead of burning the weed, it removes it mechanically with the pincer. Again the method is non-toxic.

“The disadvantage of removing weeds mechanically instead of burning it away, is that it disturbs the soil and alters the environment of the seeds produced by the weeds. Some seeds are triggered to germinate because they are moved closer to the surface or dug deeper so that the new moisture and light conditions suit the seed. By weeding without touching the soil, you can make do with less weeding. On the other hand, the burning method is more expensive,” says Frank Poulsen.

The pincer robot is ready to go on sale this year, and Frank Poulsen’s German dealer, Kress, has recently shown a film of the robot at the Dutch agricultural fair BioVak.

“My German dealer has a strong demand for these robots. I also have requests from Australia and New Zealand,” says Frank Poulsen.

Small Robot - big perspective

It was a Danish fruit grower who sowed the seeds for a fruit-picking robot which can overcome the problem of getting seasonal workers in the sector

By Nadia Louise Kristensen

There was a time when fruit pickers streamed from Eastern Europe to plantations in Western Europe each autumn to pick the ripened fruit. But now, fewer and fewer come in search of this tough and tedious work.

“We have started having problems getting seasonal workers. And in the long term, we can see that there are problems of both getting labour and competing on equal terms with those countries which have cheaper labour,” says Jan Jæger, who owns the Danish fruit plantation Kærsbo Frugtplantage.

The plantation has an annual production of between 200 and 300 tons of fruit – mostly apples and pears. And it is not the only the plantation that has that difficulty. Despite the financial crisis, problems getting labour in the future are still likely to be felt in several western countries.

“We are facing a general demographic challenge in the western world, with more and more elderly in the population. Labour will be in great demand, and the work needed to be done goes in the direction of job types that require human understanding. So it will be necessary to use robots for work that is dull, dirty and dangerous. That is where we must release human labour from, initially,” opines Christian Klit Johansen, project manager for the apple-picking robot at RoboCluster, a Danish innovation network which is engaged in developing robots, automation and intelligent mechanical systems.

That is also the reason why Robo-Cluster seized on the idea when Jan Jæger contacted them three years ago
for assistance in developing a robot which could replace the declining labour pool in Denmark and other western countries.

But the challenges are considerable. To put it mildly, trees are not robot-friendly. Branches stick out in all directions from different places on the trunk. And then there is the problem of assessing whether the apple is ripe for picking.

“When you pick an apple manually, you make a picking movement where you can feel whether the stalk is ready to release the fruit. How do you replicate that feeling in a robot?” asks Christian Klit Johansen.

One also has to ensure that the robot does not pull so hard that it bruises the fruit.

The research team cannot yet disclose how they have solved the problem since they are in the process of filing for a patent on the special robot.

Robots must match people

The researchers have made the first tests, and the robot is picking as fast as a human being, but in contrast to human beings it does not need rest periods. So the harvest is likely to be gathered faster.

“Experience from industry shows that if you present robot solutions that do the job just as well as people, but slower, then industry does not accept it – even if it’s cheaper. So the robot must be at least as quick as a human being if fruit growers worldwide are to accept the apple-picking robot,” says Christian Klit Johansen.

The robot has no limitations linguistically and can in principle function in any country, but China for example is however still excluded as a potential buyer.

“The target group is countries with wages on a level with Denmark, for example Germany, France, Italy and USA. China is not so interesting as long as it has such inexpensive labour,” he says.

The earliest the robot will be completed is in three years. There are some adjustments to be made before it is ready to go into plantations around the world.

“We have reached a point where we have all the building blocks, but there are always some surprises that will surface, which take time to handle,” says Christian Klit Johansen.

In the long term, the aim is to develop the robot so that it can help to thin out the fruit trees in spring and prune branches in winter.

At Kærsbo Frugtplantage, they are looking forward to the completion of the robot.

“We will be able to compete better because we do not need to use expensive labour, and we will also eliminate the uncertainty of whether we can get enough seasonal workers for the harvest. But it is also a major incentive that we will eliminate monotonous work for human beings” says Jan Jæger.