This article gives comprehensive information about automated guided vehicles. Read further to learn more about:
- What are automated guided vehicles?
- Common types and applications
- Overview of an AGV navigation system
- AGV Locomotion
- And much more…
Chapter 1: What are Automated Guided Vehicles?
Automated guided vehicles (AGV) or mobile robots are types of guided robotic systems that are not bounded by a fixed range of motion. Rather, it is self-contained and can move along a line, surface, or space. It differs from a robotic arm which is attached to a base and is composed of links and joints. In most instances, automated guided vehicles and robotic arms are combined. Automated guided vehicles serve as the mobile platform where robotic arms are attached to perform more versatile functions such as remote handling, telemanipulation, scanning, probing, and so on. AGVs are used in a wide range of applications such as manufacturing, warehousing, inspection, exploration, transportation, and military.
AGV systems are part of a broad branch of automation. The design of their controls is more complex than robotic arms. Aside from dealing with locomotion, there are aspects of navigation such as perception, localization, path planning, and motion control. Robotic arms usually operate in a controlled environment where their trajectory is already known or easily predicted. AGVs, on the other hand, must be able to traverse through the environment where initially unknown restrictions are present.
Chapter 2: Common Types and Industrial Applications of AGVs
AGVs are generally used in logistics. Other functions such as exploration, inspection, and service robotics comprise only a fraction of the industry. Thus, automated guided vehicles are categorized according to their load and method of transport.
Forklift AGV: Forklift automated guided vehicles are simply an automated guided vehicle navigation system integrated into a forklift. They are suitable for floor-level pallet pick-up and can stack pallets at various heights. Forklift AGVs are widely used in automatic storage and sorting systems, particularly automatic warehouse racking. The navigation system can be overridden to allow manual control.
Underride AGV: Underride automated guided vehicles or automated guided cart (AGC) is a type of AGV that lifts the load by driving underneath a basket or cart and lifts it slightly. At the destination, underride AGVs can orient and drop the load without intervention. They are mostly used in hospitals for delivering food, linens, and medical supplies.
Towing AGV: Towing or tugger automated guided vehicles are used to pull undriven carriers or trailers. Since load-carrying does not involve lifting, it can handle multiple loads in contrast with forklift and underride AGVs. However, they are solely for transport and cannot position the loads to their location.
Unit Load AGV: Unit load automated guided vehicles are designed to transport unitized or palletized goods. This type does not lift the load off from the floor which requires other lifting equipment such as conveyors, cranes, or forklifts for loading and unloading.
Assembly AGV: Assembly automated guided vehicles are used for transporting goods through an assembly process. Since the assembly process is generally a controlled environment, assembly AGV navigation is much simpler than other types. Their driving speeds are also lower in comparison. Their locomotion system is highly maneuverable that enables them to fit and orient accordingly to the assembly stations.
Heavy Load AGV: Heavy load automated guided vehicles are widely used in paper and steel mills where rolls of finished products are transported for storage or distribution. They have more robust construction than the other types and are equipped with more safety equipment.
Mini AGV (Small Load Carriers): Mini automated guided vehicles, also known as small load carriers, are designed to transport small parts or objects and usually work in swarms or fleets. They can move at high speeds and with great flexibility. Their locomotion system is usually composed of a three-wheel drive system for stability and minimal turning radius. They are commonly used in high-selectivity racking systems.
Chapter 3: The AGV Navigation System
Navigation is the ability of the guided vehicle or mobile robot to determine its location and know autonomously the direction of where it should proceed while avoiding collisions and unsafe conditions. Navigation can be divided into four fundamental components: perception, localization, path planning, and motion control.
The concept of acquiring data for mobile robot navigation is broader and more complex than robotic arms. The perception of mobile robots, similar to robotic arms, is achieved using sensors. Its higher sophistication arises from its ability to measure and relate its position globally or over a wide range.
Sensors can be classified according to two functional axes. First is the classification of where the measurement was taken. These are proprioceptive and exteroceptive sensors. Proprioceptive sensors measure internal parameters such as motor speed, load, temperature, system voltage, and current. These types of sensors are typical with robotic arms as well. Exteroceptive sensors measure parameters from the environment such as distance, electromagnetic wave intensity, and acoustic amplitude.
The second functional axis is according to whether the sensor absorbs or emits energy from the environment. These are known as passive or active sensors. Passive sensors measure physical parameters from the environment by absorbing energy such as loads, electromagnetic waves, or vibration. Active sensors, on the other hand, perform measurements by first emitting energy into the environment and analyzing its reaction. An example is an active sonar which involves beaming an acoustic wave and analyzing its echo.
Sensors for AGVs are summarized below.
Tactile Sensors: This includes contact switches and proximity sensors. The method of measurement can be mechanically or through physical contact (limit switches) or other physical phenomena such as magnetism (reed and hall effect switches) and electric induction (inductive switches).
Heading Sensors: This includes compasses and gyroscopes. Heading sensors are used to determine the orientation of the robot relative to a fixed, external reference point or frame.
Wheel and Motor Sensors: Wheel and motor sensors are used to measure the angular position, speed, and acceleration of a motor or wheel. An example of this is an encoder in a servo motor where the feedback signal is used to control the motor drive.
Motion and Speed Sensors: These sensors measure the speed of the robot relative to a fixed or moving object. These sensors are exteroceptive, in contrast with proprioceptive wheel and motor sensors.
Acceleration Sensors: These are sensors used to determine the acceleration of the robot. Most of the time, acceleration is a less important value. Position can be indirectly determined from acceleration, initial position, and orientation through dead reckoning. A combination of acceleration and heading sensors are commonly referred to as the Inertial Measurement Unit (IMU).
Beacon-based Sensors: These sensors use a known fixed reference point or frame to determine a robot’s position and orientation. An example is the global navigation satellite system (GNSS) where robots use an electronic receiver that receives orbital data and compares it with the time-of-flight measured from three or more satellites to calculate its position and orientation.
Active Ranging Sensors: Active ranging sensors are sensors capable of transmitting and receiving signals. A signal is radiated towards an object or reference point which reflects part of the signal which is then measured and analyzed using concepts of reflectivity, time-of-flight, and triangulation. Example of an active ranging sensor is lidar, radar, and sonar.
Visual Sensors: Vision is a high-level feature that allows the robot to analyze captured images to determine their localization. It can also perform additional functions such as obstacle avoidance and object recognition.
After gathering information either from the environment or from a fixed reference frame, the robot evaluates the data to determine its position and orientation relative to the environment through a process known as localization. A robot’s position and orientation can be determined through odometry (dead reckoning) or triangulation from fixed reference frames. In most cases, this is not enough especially when high accuracy is required. The environment has unknown obstacles and restrictions that are constantly changing. In addition, the sensors and effectors have their issues regarding accuracy and precision. To achieve full autonomy and to proceed with the succeeding steps of navigation, mapping is done to create a model of the environment that the robot can use to determine not only its location and orientation but also its goal. The robot must also be able to update information accurately and in real-time. This process of creating and updating the map while keeping track of the objects and restrictions within it is known as Simultaneous Localization and Mapping (SLAM).
Path planning is the process of determining the sequence of actions needed to be performed by the robot to reach its destination. This is the robot’s cognition process where it analyzes the map of the environment and creates output in the form of a program or instructions. In the event that some attributes in the map changes, the robot must be able to measure these changes and adjust its actions accordingly. Moreover, the robot not only determines how to get to its target location but also optimizes its path by reducing the path length while avoiding obstacles.
In path planning, four different concepts need to be described. These are the robot geometry, the robot effectors degrees of freedom, the map of the environment, and the initial and target configurations. To solve the robot’s path planning, these four concepts must be translated into what is known as the configuration space. in the configuration space, the possible configurations of the robot and the space occupied by the obstacles are represented. The robot in the configuration space can be represented as a point defined by coordinate vectors instead of a rigid body. By reducing the robot to a point, the obstacles are somehow inflated by the size of the robot to compensate. Knowing the possible configurations of all objects in the map, a robot’s trajectory can be determined which corresponds to a continuous curve or path.
Motion control is the ability of the robot to execute its planned or programmed sequence of actions by feeding input signals to its drivers, actuators, and effectors. For mobile robots, the control system is typically a closed loop. The most common closed-loop control used in robotics is Proportional-Integral-Derivative (PID) control which is a form of feedback control. Feedback control allows the robot to correct any disturbances or errors to its trajectory by continuously measuring parameters internally and externally. Using a PID controller mathematically represents the error signal and the proportional, integral, and derivative gains. These three factors allow the controller to quickly eliminate the error while maintaining a stable signal and avoiding overshoot.
Chapter 4: Types of AGV Navigation Systems
Integrating the processes of perception, localization, path planning, and motion control creates the whole navigation system. There are different types of navigation systems that can be created with the combinations of sensors, controllers, programs, and algorithms. Enumerated below are the most commonly used navigation systems for automated guided vehicles.
Physical Guides: Physical guides include guide tracks, tapes, and wires that are detected either actively or passively. This type of navigation system uses fixed reference points or environment landmarks which are measured and evaluated by sensors and controllers. Since the automated guided vehicle will rely on predetermined paths for navigation, the process of path planning can be preprogrammed to its system.
An example is an inductive guide track or wire guidance system which is made of a current-carrying conductor embedded into the ground or floor. To sectionalize the tracks, it is divided into segments that can be turned on or off. An alternating current flows through the wire which generates electromagnetic waves that can be detected by the mounted sensors at the bottom of the automated guided vehicle. The sensor is composed of two coils. Currents are induced through these coils which become analog signals fed to the feedback controller.
Other examples of physical guides are magnetic, metallic, and optical guide strips. These strips are mounted on the surface of the floor which is detected by magnetic, inductive, or optical proximity sensors at the bottom of the AGV. Magnetic proximity sensors operate through a physical phenomenon called the Hall-effect which allows them to detect magnetic materials. Inductive proximity sensors, on the other hand, are also active sensors that utilize the principle of electromagnetic induction to detect metallic materials. Optical sensors determine the path by detecting the recognizable features (color) of the tape. Physical guides are less expensive than wire guidance systems and can easily be reconfigured. However, they are not suitable in dirty and high traffic areas.
Anchoring Points: Anchoring points are also physical guides, but it allows free navigation. Instead of installing a predetermined path using wires and tapes, a grid of permanent magnets (magnetic bars) or transponders are set on the floor to determine the location and bearing of the automated guided vehicle. Like the magnetic strip guide, magnetic proximity sensors are also mounted on the bottom of the AGV. The path followed by the robot is based on preprogramming or path planning.
Laser Navigation: Laser navigation is a free navigation system that uses active ranging light sensors for localization. Markers such as reflective foils or tapes are mounted on walls or objects. These markers can be easily detected by the laser sensor. A minimum of three markers is needed to enable triangulation. Since localization and path planning is performed by the automated guided vehicle, the trajectory is highly flexible. The optimum path can also be computed.
Global Positioning System (GPS): GPS navigation is used outdoors where artificial markers are impractical to be installed. The GPS satellite acts as a beacon that sends data to the AGV which is then used to triangulate its position. The downside of solely relying on GPS is its low accuracy especially indoors. To maintain a reliable signal, there must be a clear line of sight between the satellite and the automated guided vehicle.
Chapter 5: AGV Locomotion
Locomotion is the ability of automated guided vehicles to propel themselves from one point to another. A common mode of locomotion is by the use of wheels. Wheels are widely used due to their simplicity and minimal friction loss. However, there are certain limitations for wheeled automated guided vehicles in environments with irregular and uneven surfaces. This is where legged robots are usually preferred. Other modes of locomotion exist which usually mimic movements found in biology such as walking, crawling, and sliding.
Wheeled AGVs: The wheel is a man-made technology that is used due to its high efficiency and simple mechanical implementation. They can be designed to have a wide degree of freedom, high stability, and excellent maneuverability. In the field of robotics, wheels can be classified according to their kinematics. The different wheel types used in automated guided vehicles are enumerated below.
Standard Wheel: A standard wheel offers two degrees of freedom: one is rotation around the wheel axle, and the other is rotation around the contact point. The axis of rotation with respect to the contact point passes through the center of the wheel, usually along the plane of the wheel. This design allows steering without imparting additional forces to the robot chassis, provided that the wheels are aligned with no camber and toe.
Caster Wheel: As with the standard wheel, a caster wheel offers two degrees of freedom. One is rotation around the wheel axis while the other is around an offset from the center of the wheel. Caster wheels are generally used to provide support for the chassis. It is rarely used for maneuvering and delivering motion since steering using caster wheels exerts forces on the chassis. The main advantage of using caster wheels is its automatic alignment when moving forward after turning.
Mecanum Wheel: Sometimes referred to as Swedish wheels, a Mecanum wheel has three degrees of freedom: one is rotation around the wheel axis, another is rotation around the rollers, and last is rotation around the contact point. A Mecanum wheel has rollers along the circumference of the main wheel arranged in 45° angles. Another configuration of this type is an omnidirectional wheel with rollers oriented in 90° angles. To provide omnidirectional movement, three or more Mecanum wheels are installed to the chassis and rotated in combinations of clockwise and counterclockwise rotations.
Ball or Spherical Wheel: A spherical wheel offers three degrees of freedom or rotation around all three axes. True spherical wheels are mostly conceptual since implementing them into a large chassis brings issues such as opposing rotation between drivers, unpowered stability, and difficulties of power transmission with the drive system.
Different types of wheels can be combined and configured to produce the necessary stability and maneuverability. A statically stable robot requires a minimum of two wheels. Two-wheel stability can be achieved by lowering the robot’s center of mass below the wheel axle. For three or more wheels, stability is achieved by containing the center of mass within the polygon formed by the points of contact between the wheel and ground. In terms of maneuverability, the usual configuration is a two-wheel drive with one or two undriven wheels for steering. A differential is used for maneuvering with fixed standard wheels or wheels with only one degree of freedom. Four-wheel drives also exist which use two pairs of driven and steered standard wheels or four individually driven and steered Mecanum wheels.
- Standard Wheel: A standard wheel offers two degrees of freedom: one is rotation around the wheel axle, and the other is rotation around the contact point. The axis of rotation with respect to the contact point passes through the center of the wheel, usually along the plane of the wheel. This design allows steering without imparting additional forces to the robot chassis, provided that the wheels are aligned with no camber and toe.
- Legged AGVs: Legged AGV is another terrestrial automated guided vehicle developed to take advantage of its high maneuverability in irregular terrain. It can cross gaps or holes as long as the reach of its legs exceeds the width of the gap. However, this type is less often used in industrial applications due to limitations in load carrying capacity and the mechanical complexity of the leg assembly. A leg is composed of several links and joints which typically require independent drivers. Multiplying this with the number of legs leads to a significant increase in the robot’s weight. This decreases its load-carrying capacity. Moreover, additional actuators require more power and control. Effort in developing legged AGVs is mostly consumed by designing the kinetics and controls of the legs instead of focusing on other aspects such as navigation and localization.
Aerial AGVs: Aerial AGVs use different principles of flight. To summarize, aerial automated guided vehicles can be divided into two: Lighter Than Air (LTA) and Heavier Than Air (HTA). LTAs are balloons and blimps. Balloons are limited to elevation control while blimps have propellers to move laterally. HTA, on the other hand, are gliders, planes, and rotorcrafts. Gliders and planes rely on wings and airfoils and their dynamic reaction with air. Rotorcrafts generate lift and lateral movement using rotary blades or propellers. Among the aerial AGVs, rotorcraft is the most promising to be applied for practical use. They can be made very light, compact, and easy to control. Moreover, they can take-off and land vertically. Unmanned aerial vehicles (UAV) or drones in the form of rotorcrafts are now available for consumer use in applications such as photography, inspection, navigation, and agriculture.
- Submersible/Aquatic AGVs: The principles used in this type is similar to aerial AGVs. Submersible automated guided vehicles, also known as autonomous underwater vehicles (AUV), can be likened to blimps that float using buoyant force and propel laterally using rotor blades. They are used in scientific and industrial applications such as seafloor mapping, environmental monitoring, and pipeline and cable inspections.
- Automated guided vehicles (AGV) or mobile robots are types of guided robotic systems that are not bounded by a fixed range of motion. They are generally used in logistics.
- Navigation is the ability of the guided vehicle or mobile robot to determine its location and know autonomously the direction of where it would proceed while avoiding collisions and unsafe conditions.
- Integrating the processes of perception, localization, path planning, and motion control creates the whole navigation system. Common navigation systems are physical guides, anchoring points, laser guides, and GPS.
- Locomotion is the ability of AGVs to propel themselves from one point to another. The common mode of locomotion is through rolling components or wheels which are widely used due to their simplicity and minimal friction loss.