Revolutionizing vineyard management: 19 OXIN robots at Pernod Ricard

A fleet of 4 robots are mowing and weed spraying together. – Photos: Maxence Guillaumot
A fleet of 4 robots are mowing and weed spraying together. – Photos: Maxence Guillaumot

Pernod Ricard, a major winery located in the Marlborough region of New Zealand, has deployed a fleet of 19 OXIN machines to handle various tasks across its extensive vineyards. This implementation of automation aims to enhance efficiency and safety but comes with logistical and technical challenges inherent in managing many robots across a vast area. Future Farming decided to participate half day within their operation, to get their experiences as users and partners in the development of these machines.

5 and half years before project kick off, Pernod Ricard is now operating 19 machines in two different regions. Since 2020, they arrived on the vineyard in patches of 5. The OXIN robots are designed for multitasking, capable of mowing, herbicide spraying, trimming, and defoliating. Powered by 130hp Cummins engines and equipped with large diesel tanks, these machines can operate for extended periods without frequent refueling. Their robust design is complemented by advanced safety systems, including lidar and cameras that detect obstacles and nearby humans, ensuring operational security while minimizing the risk of accidents​​.

David Allen, the vineyard transition manager at Pernod Ricard, highlights the versatility and robustness of these machines, which are crucial for maintaining the work quality traditionally achieved by tractors. According to David, “multi-tasking is key, because there is always something to do it a vineyard, so we can keep them busy all year”. However, he also points out challenges, such as frequent stops caused by overly sensitive safety systems and connectivity issues that still impact operational efficiency​​. The oldest robot count more than 2 500h and the newest machines around 500 ha. Thanks to a good collaboration with the manufacturer, they keep improving every day.

Text continues below picture

The last major new features implemented were the double navigation called “Oxnav” which consist of using AB line as reference and lidar to adapt to adjust within the row, according to David, “a game changer in the working quality”. The second is the safety, very important for the Pernod Ricard team. Additionally, to the cameras in the front and in the back, the bumpers in the front of the tools, the machine can detect now detect any human 15m away. Finally, a machine is now working in an Australian vineyard, and David expects from this new working conditions, and David explained this new experience in tough Australian conditions will also bring new things to his machines too.

Managing efficiently a fleet of 19 robots

During a half-day site visit with a fleet of four robots, Future Farming observed the intricacies of fleet management at Pernod Ricard. To maximize efficiency, it’s crucial that the machines operate as consistently as possible, as time spent on logistics is essentially time wasted. Each robot is stationed within large 400-hectare plots, where it performs all required tasks, with operators ensuring that all necessary inputs—fuel, water, chemicals—are readily available to the machines.

Over the past four years, Pernod Ricard has refined its organizational strategies. Each robot fleet, comprising four machines, is overseen by a dedicated operator who monitors performance and a second operator who ensures supply delivery, usually for more than one fleet. By the end of the season, a total of five operators manages the full fleet of 19 robots. During peak seasons, operations may shift to a double shift schedule to accommodate increased workload.

Operational strategy

The operational strategy involves spacing the robots sufficiently apart to avoid interference and assigning specific rows to each robot. This arrangement isn’t the most efficient, as it requires operators to cover greater distances. Regular maintenance is performed weekly by the OXIN team, and remote updates are applied as needed to keep the machines running smoothly.

Control of the robots is facilitated through a single, simple, and efficient internet portal interface, which allows for monitoring up to four robots simultaneously. If a robot encounters an issue, the operator can quickly check the situation using the camera feed through their tablet, make necessary adjustments, and restart the work. Plans are in place to upgrade this interface to manage even more machines and enhance its ergonomic features. Sensors and cameras generally allow operators to quickly diagnose and resolve most issues, while remaining in their ATV.

Text continues below picture

Only user interface for the operator sitting in his ATV
Only user interface for the operator sitting in his ATV.

Opportunities for efficiency improvements

During our visit to Pernod Ricard, one machine encountered some connectivity issues despite the presence of two Starlink systems around the field. It appears that operating four machines in proximity might be contributing to these disruptions. Conversations with an operator who has been with the project for 15 months revealed that while there has been significant improvement, the machines still experience about 20 minor stoppages per day, with better days seeing only 2 to 3 interruptions per machine.

A primary cause of these stoppages is the safety protocols of the machines; the lidars occasionally misidentify vines or tall grass as humans. Although there have been advancements in human recognition technology, further refinement is needed to reduce “false positives”. Additionally, the current limitations of the remote-control system require operators to physically cross the field to address issues when they are beyond a certain range.

Tool changeovers could also present a challenge, currently taking half a day to switch implements. Moreover, there is a global lack of fleet optimization; the operational strategy requires leaving 15 rows between working machines, complicating logistics and necessitating the operator to cross the entire plot for an intervention. The U-turn maneuverability of the machines is also way slower than what is achievable with a good tractor driver. These examples could enhance operational efficiency.

Text continues below picture

OXIN machine with a mower and weed spraying equiped.
OXIN machine with a mower and weed spraying equiped.

Long term profitability vision

Although the fuel consumption of the robots is comparable to that of traditional tractors, these machines have the advantage of performing multiple tasks simultaneously. However, the speed of operations can be hindered when slower tasks dictate the pace, slightly reducing overall productivity, like mulching which is around 3km/h.

To improve efficiency, David is considering the possibility of enabling the robots to handle up to three tasks at once, thereby maximizing labor and fuel efficiency. He recognizes the challenge in matching the productivity of conventional tractors, especially since the robots currently cannot manage two rows at a time (like a trimmer) without potential safety risks. However, he believes that deploying them in larger fleets—potentially six robots managed by just one operator—could significantly accelerate profitability beyond initial projections.

From an environmental perspective, David is also focused on reducing the vineyard’s carbon footprint. By decreasing the number of machines passes and the associated soil compaction, not only does this improve the soil health, but it also contributes to lower CO2 emissions.

Text continues below picture

Rear of the machine after mowing.
Rear of the machine after mowing.

Join 17,000+ subscribers

Subscribe to our newsletter to stay updated about all the need-to-know content in the agricultural sector, two times a week.

Maxence Guillaumot Product and Market Analyst, AgTech Market
More about