Elegance in Design Prize - 2023 Farm Robotics Challenge

Our team received the Elegance in Design Prize at the inaugural Farm Robotics Challenge, hosted by the AI Institute for Next Generation Food Systems (AIFS), farm-ng, and UC ANR, for Robo-ag, an autonomous targeted pesticide application system.

Robo-ag in Vineyard Spray Map UI
Left: Robo-ag robot in a vineyard trial. Right: The real-time targeted spray control interface.

Background

Resource overuse is a significant challenge in modern agriculture. Traditional “blanket” spraying methods lead to:

  • 70 million pounds of pesticide active ingredients wasted annually in the U.S.
  • $1.4 billion in economic losses due to incorrect application.
  • Environmental damage including eutrophication, harmful algal blooms, and soil acidification.

In our target 10-acre vineyard in Napa County, we aimed to transition from broad application to precise, targeted control to save costs and protect the environment.

Targeted spray system

Using the farm-ng Amiga robotic platform, we developed an autonomous spray applicator capable of delivering precise doses to specific plants.

The system has three main parts:

  • GPS-Guided Path Following: Uses onboard sensors and a line-following algorithm to navigate vineyard rows.
  • Dynamic Spray Control: A self-pressurizing tank system with solenoid-controlled nozzles that activate only at defined points.
  • Safety: Integrated weather shields and automated system checks (GPS, pressure, battery) to ensure reliable operation.

How it works

The robot’s operation follows a task hierarchy:

  1. Start & Test: Remote activation and automated subsystem diagnostics.
  2. Navigate: The Amiga uses GPS and line-following to maintain alignment within vineyard rows.
  3. Targeted Spray: The system cross-references real-time GPS coordinates with a pre-defined application map (JSON-based) to trigger the spray nozzles.
Robo-ag System Architecture
System architecture diagram showing the integration of GPS, control computer, and spray system.
Robo-ag Workflow Task Hierarchy
Detailed task hierarchy for the autonomous spraying process.

GPS Navigation & Data Mapping

To achieve high precision, we used GeoJSON-based application maps. The system tracks the robot’s current position and cross-references it with pre-defined spray points.

GeoJSON Spray Points
GeoJSON configuration for target spray points, mapping geolocated coordinates to application conditions.
GPS Trajectory
Visualized GPS trajectory of the Amiga robot during a vine-row trial, showing autonomous navigation performance.

Expected benefits

By moving from blanket application to targeted spraying, our system can:

  • Reduce herbicide and fertilizer usage by up to 90% in many scenarios.
  • Minimize manual labor required for scouting and application.
  • Improve crop health by ensuring optimal nutrient delivery without over-exposure.