Robots in Precision Agriculture
Robot System Components
Robot systems are characterized as programmable mechanical devices that interact with their environment, including people, using numerous sensors, actuators, and human interfaces to perform a specific task. Agricultural robots are typically built to perform various tasks, such as planting, weeding, pruning, picking, harvesting, packing, and handling. To perform the required task, a robotic system used in agricultural and field environments generally consists of some or all of the following components: (1) a vision system to detect and localize target objects and obstacles that may be on the way to get to the objects (e.g., fruit and branches in robotic apple harvesting); (2) data-logging device for their further processing; (3) a manipulation and end-effector system for reaching and engaging with the target objects; (4) a path planning and control system for robotic manipulation; (5) pre- or post-manipulation object handling system (e.g., a conveyance system to bring harvested fruit onto a container/bin); (6) a holding platform with navigation and guidance system for mobility; and (7) a surround awareness system for operator safety.
Robot Vision Systems
In robotic systems, the vision system serves as the robot’s "eyes", allowing it to perceive, interpret, and interact with its environment. This is especially critical in precision agriculture, manufacturing, autonomous vehicles, and service robotics, where tasks depend on recognition, measurement, alignment, and navigation. A robotic vision system refers to a combination of hardware and software that enables a robot to capture and process visual information to make decisions or perform actions. Vision systems in agricultural robots are crucial for enabling autonomous decision-making, navigation, object detection, crop health monitoring, and precision tasks. The core components of these systems include the following:
Capabilities of Vision Systems
Vision systems in agricultural robotics enable robots to "see" and interpret their environment, allowing for tasks like crop monitoring, harvesting, and automated weeding. They use cameras and image processing algorithms to identify objects, assess plant health, and guide robotic arms for precise actions.
Machine Vision Algorithms
Image segmentation, object detection, and image matching are the keys to the machine vision navigation system of an agricultural robot. Image segmentation algorithms are used to segment an image into different are-as, such as plants, land, etc., and the quality of image segmentation is an important factor affecting the ex-traction effect of the navigation line. In addition, due to the existence of obstacles in the agricultural environment, agricultural robots without obstacle avoidance function have caused a high accident rate.
Data Logging
Data logging in robotic systems involves capturing, storing, and displaying data generated by the robot's sensors and other components. This data can be used to analyze the robot's performance, identify patterns, and even predict future events. It's a crucial aspect of industrial robots, enabling recovery from failures and supporting emerging AI/ML capabilities like predictive maintenance. Data logging is important for several reasons:
Manipulation Systems
In robotics, manipulation refers to a robot’s ability to physically interact with objects in its environment—grasping, moving, assembling, or modifying them. Robotic manipulation involves three critical components:
Robotic Arms
As a core component of robots and the broader field of robotics, robotic arms replicate human arm movements and are employed to replace manual labor in agricultural scenarios. A robotic arm is a mechanical arm designed to perform tasks that mimic the movements and functions of a human arm. It consists of links (segments) connected by joints, allowing for a range of motion and the ability to grasp and manipulate objects. Robotic arms have a flexible arm comprising several joints, allowing them to move in multiple directions and reach a wide range of positions.
Robot Actuators
Robot actuators are the “muscles” of a robot, the parts that convert stored energy into movement. They are an integral part of any robotic system. Actuators are typically powered by air, electricity, or liquids. The type of actuator used can greatly affect the robot’s performance and efficiency. In the field of robotics, actuators play a crucial role. They are responsible for making the robot move, whether it’s a simple movement like the rotation of a joint or a more complex one like walking or grabbing objects. Their versatility allows for various applications, from industrial automation to sophisticated humanoid robots. Without actuators, robots would be static and incapable of any movement or action.
End Effectors
The end-effector, also known as an end-of-arm tool (EOAT), is a special tool at the end of the robot's arm, giving the robot its hands and fingers. It allows a robot to pick up, manipulate, and work with objects like your hand would. At present, a wide variety of end effectors have been developed, with fingers, attractors, needles, spray nozzles, scissors, and robotic arms, to grip, cut, attach, or press into crops to effectively perform all biological production processes, which include picking, harvesting, spraying, sowing, transplanting, shaping, and primary processing. Because end-effectors come into direct contact with the objects being manipulated (or being operated on), the end-effector efficiency greatly determines the performance of a robotic system.
Path Planning
Finding a continuous route for the robot to travel from the initial state to the target state/configuration is called path planning. The mobile system uses a known map of the environment stored in the robot’s memory to perform path planning. The state/configuration provides the robot with a possible position in the environment, and it can move from one state/configuration to another by performing various actions. Path planning is a crucial aspect of robot control and must be collision-free, reachable, smooth, safe, predictable, and responsive to enable robot integration in industry and society. Path planning algorithms are used by mobile robots to determine safe, effective, collision-free, and cost-efficient paths from the starting point to the destination. A few popular categories of path-planning algorithms for autonomous vehicles include:
Pre- or Post-Product Handling
A conveyance system in robots agriculture, used for pre- or post-manipulation object handling, is a crucial component for efficiently moving harvested fruit onto containers or bins. This system, often in the form of a conveyor belt or a robotic arm, ensures smooth and organized handling of the produce, minimizing damage and optimizing workflow.
Holding Platforms
Robots on holding platforms, also known as handling systems, are integral to precision agriculture. These platforms can be stationary, mobile, or collaborative, and are equipped with various "end-of-arm-tooling" (EOAT) to grasp, carry, and place objects. They increase efficiency, reduce manual labor, and enhance safety by automating repetitive or hazardous tasks. Today’s robots can generally be grouped into four categories.
Autonomous Mobile Robots (AMRs)
An autonomous mobile robot is a type of robot that can understand and move through its environment independently. AMRs use a sophisticated set of sensors, artificial intelligence, machine learning, and compute for path planning to interpret and navigate through their environment, untethered from wired power. Because AMRs are equipped with cameras and sensors, if they experience an unexpected obstacle while navigating their environments, such as a fallen box or a crowd of people, they will use a navigation technique like collision avoidance to slow, stop, or reroute their path around the object and then continue with their task.
Automated Guided Vehicles (AGVs)
While AMRs traverse environments freely, AGVs rely on tracks or predefined paths and often require operator oversight. These are commonly used to deliver materials and move items in controlled environments such as greenhouses and factory floors.
Articulated Robots (Robotic Arms)
An Articulated robotic arm is a type of industrial robot designed like a human arm with different segments known as links. These links are usually connected by joints that give this robotic arm more flexibility and precision. They can either be rotary joints (revolute) or linear (prismatic joint). Generally, these Industrial robotic arms have around two to six links (sometimes more) that grant them a certain degree of freedom (DOF) for movement. The most common configurations are 4-DOF and 6-DOF, with the latter having more flexibility. AI allows articulated robots to perform tasks faster and more accurately. AI technologies infer information from vision sensors, such as 2D/3D cameras, to segment and understand scenes and detect and classify objects.
Cobots
Cobots are the latest technology in robotics and have changed the world of automation significantly. The name cobot is a derivative of “collaborative robot.” These robots are collaborative because they can safely work together with people. The application scenarios of AI and machine vision in collaborative robots are gradually expanding, with increasing penetration rates. Machine vision can assist collaborative robots in more accurately identifying and tracking targets. Combined with artificial intelligence decision-making capabilities, collaborative robots can quickly learn and optimize methods for task execution, achieving higher efficiency in task completion.
Surround Awareness Systems
Surround awareness systems for robotic platforms are designed to provide robots with a comprehensive understanding of their environment, enabling them to navigate, interact, and perform tasks safely and effectively. Surround awareness, also known as situational awareness, involves a robot's ability to perceive, comprehend, and project the status of its surroundings in the near future. This allows the robot to make informed decisions and respond appropriately to changes in its environment.
Sensors
For robot navigation and achieving surround awareness, several types of sensors are employed, each with its strengths and weaknesses, contributing to a comprehensive understanding of the environment. A breakdown of common sensor types includes the following:
Data Processing and Interpretation
Robots use computer vision algorithms to analyze the data from cameras and other sensors. These algorithms can perform tasks like object detection, image segmentation, and feature extraction. Combining data from multiple sensors (i.e., sensor fusion) enhances a robot's understanding of its surroundings.
Click on the following topics for more information on robots in precision agriculture.
