Research · Current project
Olive Picker
An autonomous robotic platform for olive harvesting, targeting the underserved gap between handheld electric harvesters and industrial trunk shakers. V1 is a bench-top R&D platform; V2 is targeted as a field-deployable harvester for small and mid-sized Greek olive farms.
The problem
The Greek olive harvest depends on two extremes. Handheld electric harvesters are reliable but require a labour team — typically 4–8 people working full days. Industrial trunk shakers automate the work but require capital scale and uniform orchards.
The vast middle — small and mid-sized farms of 5 to 50 stremmata (1.25 to 12.5 hectares) — has no good autonomous option. Labour shortages, narrow ripeness windows, and rising input costs make this gap increasingly painful for the farmers who tend the majority of Greece's olive groves.
Our approach
V1 is a bench-top robotic platform built around a deliberate set of design decisions, each documented as it was made. Four are central.
-
Branch processing, not single-olive picking.
Selective single-olive picking is still an unsolved research problem at production scale globally. Branch processing — sweeping a comb head along a branch to detach olives — is the proven mechanism used by every commercial handheld harvester. We adopt it directly rather than betting on unsolved research.
-
Commercial whipping head, adapted to a robot.
Rather than design a stripping mechanism from scratch, V1 uses a commercial handheld harvester head (Volpi Olytech Dualcomb 255SP or Zanon Karbonium Eco Genius class) mounted on the robot via a custom flange adapter. Field-proven mechanism, modular, easily replaceable.
-
Arm-as-positioner.
The robotic arm provides only end-effector positioning. All task-specific motion — the high-frequency whipping action — lives in the head, which has its own motor. The arm-and-head interface, both mechanical and electrical, is standardised so different heads can be swapped without changes to the arm or its control software.
-
Wide passive catch.
Detached olives drop into a wide passive catch area below the working envelope, separate from the picker. This removes the need for a moving catch system that follows the arm and keeps the end-effector light.
V1 architecture
The bench-top platform combines a six-axis positioning arm with a self-contained whipping head and a wide passive catch system. The schematic below shows the head — the most novel part of the integration — viewed from the side, with the branch entering from the left.
- Robotic arm
- UFactory xArm 6, 700 mm reach, 5 kg payload, 6 DoF, ROS 2-compatible
- Vision
- Intel RealSense D455 + D435i depth cameras
- End-effector
- Commercial whipping head (Volpi Olytech Dualcomb 255SP or Zanon Karbonium Eco Genius 12V), custom CNC flange adapter
- Frame
- 4040 aluminium extrusion, 1.5 × 1.2 × 2 m bench
- Catch system
- Wide passive catch, separated from the picker assembly
- Power
- 12V DC bench supply for the whipping head
- Compute
- Linux workstation running ROS 2
From V1 to V2
V1 — bench-top R&D system (current). Goal: validate branch presentation, head positioning, and the whipping mechanism in a controlled indoor environment. Target completion: late 2026 / early 2027.
V2 — field-deployable system. Targets the 5–50 stremmata cohort. Architecture moves from a fixed bench to a mobile platform, with a deployable canopy catch system that adapts to tree shape. Build begins after V1 validates the core mechanism.
Project status
- Phase
- Pre-procurement / early V1 build
- Design decisions
- 13 documented architectural decisions
- Hardware procurement
- Quotes being collected from EU suppliers
- Last update
- June 2026