Michael Back and his son Simon, owners of the Backsberg wine estate between Paarl and Franschhoek in the Western Cape, had a problem at the time of the blueberry harvest: their pickers had to walk up to 150 m to and from the control centers of quality to deliver the fruit. .

The inefficiency of the system was obvious, so the Backs set out to find a way to reduce walking time and increase hours spent collecting and sorting.

Simon explains that they did not want to use tractors, as the rows in their garden are narrow (2 m to 2,5 m depending on the variety), and the tractors would have to pass past agricultural workers, whose number increases to more than 300 in the high season from August. to November.

Apart from this, tractors consume a lot of fuel, have a high carbon footprint, increase the risk of compaction and require drivers.

“We were looking for a solution similar to a conveyor belt where workers could place their buckets [of fruit] right where they were working. Then my father came up with the idea of ​​having a robotic transport platform that would automatically follow the collection teams up and down the lines to collect and deliver their buckets to quality control centers, ”says Simon .

Carl Malherbe, one of the farm managers, suggested that the Backs consult electronics engineer Cobus Meyer to help them develop this solution. Meyer, in turn, approached Chris von Wielligh, a fellow electronics engineer, for help with this challenge.

Since then Cobus, Chris, Michael and Simon have founded ARCi Technologies and developed what can probably be considered South Africa's first semi-autonomous agricultural robot, ARCi. The name is derived from the term 'agricultural robotic carrier' and is pronounced 'Archee' to give the machine some personality.

The first two prototypes

The original model was an old remote-controlled car from Cobus's childhood, which they adapted with an inexpensive camera to test the first version of their autonomous guidance technology.

"We wanted to show that we could implement a basic autonomous system with inexpensive parts that could guide our car down the row of a vegetable garden," he says.

The results were promising and led to the development of their second prototype, which they tested during the Backs' 2020 harvesting season. This machine, substantially larger than the first, could transport four collection boxes in one layer, as well as advanced camera technology to improve the precision of the autonomous guide system.

“The second prototype has been very successful in moving between the rows of the orchards, without accidents or injuries. However, work is still needed to improve navigation once the robot leaves the ranks, and we are addressing this with our [third] pre-commercial prototype. Outside of the ranks, we use a combination of high-precision GPS, wheel sensors, and machine learning technology to help us navigate through unfamiliar environments safely, ”Cobus explains.

The robot is considered semi-autonomous as it cannot be transported between orchards and to the battery charging station. The pre-commercial version weighs around 100kg and is 1,2m long, 0,9m wide and approximately 0,8m high.

According to Simon, the workers were concerned about the robot at first, but soon realized that it was there to help them and not to replace them.

“Berry orchards need to be harvested multiple times to ensure that the fruit is harvested at optimum maturity. Our workers are paid a base salary and the bonuses are tied to productivity, so the robot allows them to increase their profits by giving them more time to pick berries rather than having to walk [to and from the quality control center]. This, in turn, has a positive impact on the farm's income, as it results in better and larger packaging, ”says Simon.

Workers interact with the robot through a simple three-button interface located on each side. They can stop the robot and send it forward or backward on its path.

Challenges and updates

During testing, the second prototype was occasionally lost, but this was where the rows of plants did not conform to the conventional design; for example, when more than two neighboring plants were missing in a row.

However, according to Michael, this is not a significant setback, as these plants should, in any case, be replaced. He advises farmers who are considering using robotics to design their orchards so that the machine can be easily moved.

"While the terrain can be undulating, it should not be covered with grass that could obstruct [the robot's] sensors, nor should it be full of holes that could affect the movement of the machine."

Another challenge was finding a way to identify pickers' yields after they had delivered their berries. This was solved by connecting a weight scanner to the robot; Workers scan each cube upon delivery for identification purposes.

"This was not a major change, as the [weight scan] function just moved from the QA center to the robot," says Simon.

While no injuries or damage to plants were sustained during the test, some berry cubes were sacrificed in the pursuit of innovation.

Cobus explains that during the tests, the robot platform was equipped with multiple shelves to increase the volume of berries it could carry per load.

“We managed to carry loads of up to 70 kg before having stability problems caused by poor weight distribution and mechanical design. The robot is [really] capable of transporting much more [than that] ”.

The robot also carried empty buckets that workers exchanged for full ones so they didn't have to walk far to do so.

The company's third prototype, or pre-commercial version, is a significant advance over the second prototype. Not only does it feature more advanced technology, it also has a lower center of gravity and is much more robust than the previous model, making it less prone to tipping over under the weight of a heavy load.

In addition, it has been equipped with larger wheels to prevent it from slipping or sticking on wet or uneven terrain, and has a touchscreen user interface to give trained operators a more detailed view of machine status, including level. battery.

The robot is electrically powered, with the second prototype capable of working for up to two days before its battery needs recharging.

"Ideally, the robot would be recharged with a renewable energy source, such as solar, wind or hydroelectric, which would help reduce a farm's carbon footprint," says Cobus.

Unfortunately, most of the robot's electronic components have to be imported, but the body and most mechanical parts are manufactured locally, and the company is trying to design the machine in such a way that it is easy for a mechanic on the farm. . repair it or quickly change its main components when something breaks. This, according to Cobus, will reduce downtime with the commercial version.

Future Applications

ARCi Technologies will host a field day in November to showcase the pre-commercial prototype.

"We hope to generate interest and get some orders to accelerate the commercialization of the product," says Cobus.

While the initial design is intended for blueberry production, the machine should also work well for other labor-intensive crops such as table grapes and strawberries. Additionally, the company has already received inquiries from the security and aviation industries on whether the technology could be adapted for its purposes.

As the robot is modular, it can be adapted for use with other applications. According to Michael, it could be outfitted with sensors that, combined with the right software, could help identify stressed patches in orchards that could be related to disease or irrigation and fertilizer problems.

"The beauty of ARCi is that it can be used under shadow nets and in tunnels, unlike drones and satellite imagery," he adds.

Sensor technology could also be used to count flowers and assess fruit maturity, resulting in better work planning as it would allow farmers to see how many pickers they would need at any given time. It would also help the commercialization of the fruit by increasing [data on] the volumes that can be delivered at specific times.

Additionally, ARCi could be outfitted with spray equipment to help with pest control, or with an ultraviolet light for a more environmentally friendly way of controlling pests and diseases.

Simon says that ARCi does not aim to replace farm workers, but to help improve their efficiency. “Globally, agricultural labor is increasingly scarce and the same is likely to be the case in South Africa as people find more attractive jobs. The wave of agrotechnology is upon us, and we can ride that wave and remain competitive, or be crushed by it. "