Two Robot Arms Could Make Sensitive Grape Stems Accessible

What the research is about

Most robotic harvesters use a single arm. The arm grasps a crop and detaches it through movement, or cuts its stem with a specially designed gripper-and-cutter tool. Research published on arXiv on September 13, 2022, argues that this approach may not be suitable for sensitive crops and cluttered environments such as vineyards.

A grape stem can be hidden by leaves, shoots, berries or other obstacles, while the space around it may be too limited for a cutter. The study therefore examines a bimanual system—meaning a robot with two arms—for human-inspired harvesting tasks.

How it works, in simple terms

The central idea is to divide the work between two arms. One arm can grasp and reposition the cluster so that its previously hidden stem becomes visible. This movement can also create a “cutting affordance”: a suitable position and enough open space for the second arm’s cutter to reach the stem.

That differs from a one-arm system, which must grasp and detach the crop with the same arm. If the stem is hidden or the cutting tool cannot fit into the available space, a single arm has limited options. With two arms, the robot can first adjust the position of the grasped cluster and then treat cutting as a separate action.

The supplied material does not describe the sensors, camera system, arm design or control software used to carry out these tasks.

Why it matters for vineyards and wineries

Grapes are difficult targets for harvesting robots. Clusters are sensitive, their shape and position vary, and their stems are often hard to see. The practical advantage of a two-arm system could be its ability to create better access before attempting a cut, rather than simply cutting from the cluster’s initially detected position.

This could support robotic harvesting in canopies where visual obstruction and tight spaces limit one-arm machines. For wineries, gentler handling may also matter because the objective is not merely to detach a cluster, but to handle sensitive fruit without unnecessary damage.

There are important caveats. The supplied information gives no figures for harvesting speed, cutting accuracy, fruit damage, operating capacity or cost. It also does not explain how the system performs across different training systems, grape varieties, lighting conditions or canopy densities. Its practical and commercial readiness therefore cannot be assessed from the available material.

What to watch next

A key next step would be evidence that both arms can cooperate reliably in a real vineyard. Important issues include detecting an obscured stem, moving the cluster without damage, planning a safe route for the cutter and maintaining a repeatable working speed.

Future tests that directly compare bimanual and single-arm harvesting would also be valuable. Useful measures would need to show whether creating access actually improves reliable cutting under cluttered vineyard conditions. The provided source does not report such comparative results, so this should currently be viewed as a promising robotics approach rather than a proven commercial harvesting system.

Source: arXiv – Viticulture & Grapevine (arXiv), “Bimanual crop manipulation for human-inspired robotic harvesting.”