Agri-OpenCore concentrated on three areas that make agricultural robots faster to build, safer to run and easier to deploy at scale. This page summarises what was delivered and what has happened since the project.

1. Reusable ROS 2 components and containers

The project released open-source ROS 2 packages and container images that directly support agri-robotics development. The AOC container base provides ready-made ROS 2 Humble and Jazzy images, with CUDA variants for GPU work and a browser-accessible desktop. Templates for ROS 2 packages and farm environments give new components a common structure, continuous integration and a development container. Components such as the fruit detector and topological navigation are packaged to run in containers.

2. Standards, human-robot safety and navigation at scale

  • Topological navigation represents a farm as a graph of places and paths, so that fleets of robots can plan routes, share one map and run row-specific behaviours across many hectares. It comes with map recording, validation and merging tools, a browser web app and a simulator for testing without hardware.
  • Runtime safety monitoring with sentor, which watches the health of a robot’s sensors and software and raises safety and autonomy heartbeats that the robot can act on.
  • Human detection and tracking from cameras and LiDAR, supported by the AGHRI dataset of people working around agricultural robots, with benchmarks and tools.
  • Measuring autonomy with an open logger that records autonomous distance, interventions and the mean distance between interventions during field trials.
  • Standards and risk assessment for the trusted use of robots on farms, including work within the BSI BS 8646 framework.

3. Digital twins and simulation

  • Generators for randomised tomato farms and glasshouses in Gazebo and Unity, compared in a TAROS 2024 paper.
  • Dynamic strawberry plant models for simulating harvesting, including detaching fruit.
  • A packhouse simulation with conveyors and people, used to test human-aware safety for robot arms.
  • Simulation models of the robot platforms used in the project, so the same software runs in simulation and on the farm.

Field trials

The University of Lincoln, and from 2026 its spinout JABAS.AI, ran joint navigation trials with Dogtooth Technologies in commercial strawberry polytunnels at Hugh Lowe Farms and Annabel’s Farm. At Annabel’s Farm a Dogtooth robot navigated autonomously for 13.5 km over seven operating days between June 2025 and August 2026. Dogtooth Technologies also ran its own harvesting trials.

Impact and legacy

  • Commercial exploitation. Navigation and fleet coordination technology developed at the University of Lincoln in the project is licensed to JABAS.AI, a University spinout incorporated in February 2026 and publicly launched in May 2026 (announcement).
  • Adoption by others. SowBot, an open agricultural robot platform, uses AOC topological navigation and sentor.
  • Follow-on infrastructure. The new HARVEST Robot Adoption Hub, led by the University of Lincoln, builds its platform-independent navigation services on the AOC ROS 2 stack, with Dogtooth Technologies and JABAS.AI among its partners.
  • Community. AOC components feature in the ROS in the Field workshop at ROSCon UK 2026 in Edinburgh.

Open by design

In line with the project’s aim of an open core that companies can build on, the platform’s core components were released as open source. Selected components were kept for targeted commercial exploitation by the partners. See Software and Resources for everything that is openly available.