Researchers at Texas A&M University have developed a groundbreaking robotic system to control weeds on farms without the use of harmful herbicides. This new method leverages a Boston Dynamics Spot quadruped robot, equipped with a flexible end effector and a propane torch, to effectively manage weed growth.

Weed infestation is a persistent issue in agriculture, as weeds compete with crops for essential nutrients, water, and sunlight. Traditional methods, such as manual removal and mechanized herbicide spraying, are either labor-intensive or environmentally damaging. The Texas A&M team’s robotic solution offers a sustainable and efficient alternative.

The robot uses a flexible end effector to adapt to environmental changes, including gravity, wind, atmospheric pressure, fuel tank pressure, and nozzle position, which can affect the flame’s coverage. According to the researchers, experiments have demonstrated that their system and algorithm can achieve over 76 percent accuracy in predicting reprojected images in real time.

The innovative system is designed to target and heat the center of weeds, effectively stunting their growth for several weeks without burning them completely. This method allows for precise weed control in their early growth stages.

The research team utilized Boston Dynamics’ Spot Mini quadruped robot, equipped with a Unitree Z1 manipulator arm. The arm, powered by Spot’s internal battery, can carry up to 3.2 pounds (2 kilograms). The system includes an Intel Realsense D435 RGB-D camera for weed detection and two FLIR Lepton 3.5 thermal cameras to monitor the flame.

The onboard computer handles perception, decision-making, and control. A propane tank supplies gas to the torch, ignited by a relay-controlled lighter. Hand-eye calibration ensures accurate alignment of the cameras, manipulator, and end-effector for effective weed flaming.

The software utilizes a YOLOv6 model, a single-stage object detection framework, to roughly locate weeds. The Spot robot plans a path to approach the weed using its software kit. Once in position, the RGB-D camera detects the weed’s center, and the system estimates the flame’s surface for precise targeting. The Unitree arm software plans the arm’s position for effective weed removal.

To test the flame estimation algorithm, researchers collected 156 pairs of images using thermal cameras in a controlled indoor environment. They classified the images based on light or strong wind conditions, developing a circular arc flame model that performed better in strong wind conditions compared to a baseline straight-line model.

Field experiments on a raised bed plot and a cotton field further validated the system. In five trials on the raised bed plot, the robot successfully detected and flamed common weeds like sunflower, ragweed, and smell melon, achieving 94.4 percent precision. The cotton field experiment demonstrated the system’s effectiveness in a real agricultural setting.

Looking ahead, the researchers plan to refine the motion of the mobile manipulator to achieve dynamic flame coverage for multiple weeds. They also aim to integrate new weed removal techniques, such as electrocution, and develop new algorithms to enhance the system’s efficiency. Their research, published in the journal arXiv, highlights the potential for this innovative technology to revolutionize weed control in agriculture, offering a sustainable and effective solution to a longstanding problem.

By Impact Lab