Technology
A reusable stack for agricultural robotics
Helinet's hardware-and-software platform is deliberately architected as a reusable stack. The Olive Picker is the first product built on it; the same components extend to weeding, spraying, pruning and other tree- and soft-fruit applications.
Perception
RGB-D vision combined with depth sensing provides real-time scene understanding. The system detects branches, fruit and surrounding obstacles, builds spatial models of the working area, and continuously refines them as the robot moves.
Motion
Six-axis robotic arms provide precise positioning. End-effectors — task-specific heads like the Olive Picker's whipping comb — connect via a standardised mechanical and electrical interface, making it possible to swap heads for different tasks (harvest, pruning, spraying) without changes to the arm or its control software.
Control
The control software is built on ROS 2, the industry-standard middleware for modern robotics, designed for outdoor field conditions: variable lighting, intermittent connectivity, mechanical vibration. Each subsystem can be tested and iterated independently.
Edge AI
Detection and decision-making run locally on-device, not in the cloud. Models for crop detection, ripeness classification, weed identification and disease diagnosis are deployed at the edge, enabling autonomous operation in fields without reliable connectivity.
Safety
Autonomous operation in real environments requires multiple layers of safety: hardware emergency-stop circuits, software watchdogs, fail-safe defaults, and behaviour boundaries enforced at the lowest control layer.
A platform that grows with the project pipeline
The same stack supports the application roadmap:
- Tree-fruit harvesting (cherries, apples, pistachios)
- Soft-fruit harvesting (tomatoes, strawberries)
- Autonomous weeding
- Precision spraying
- Pruning, fertilisation and field scouting (longer term)