Mobile robot systems: Navigation and localization

For seamless operation, understanding how AGVs localize and navigate is crucial. Here is an overview of the various methods and technologies used.

Localization and navigation: What’s the difference?

Localization is essential for mobile robots to operate within a system, determining where they are on the shop floor and possible routes. Navigation, on the other hand, plans the path and time needed for an AGV’s journey based on this localized data.

What types of localization are available?

AGVs can utilize various sensor technologies for localization, including:

• Magnetic sensor localization
• Camera-based localization
• Contour-based localization using lasers
• Triangulation

Each method offers unique advantages and, depending on the specific needs and applications, multiple localization technologies may be used concurrently.

Odometry: The fundamental localization method

Odometry is often used as a fundamental localization method for Automated Guided Vehicles (AGVs). This technology relies on tracking the vehicle’s direction and speed from a known starting point to estimate its current location.

Known as the dead reckoning method, odometry uses the starting point as a reference to calculate the AGV’s current position continuously.

While odometry provides essential basic data, it primarily serves a supportive role, enhncing the accuracy and reliability of other, more advanced localization technologies.

Magnetic localization

Magnetic localization is notably straightforward to integrate. In this method, a magnetic track is affixed to the floor. Sensors positioned beneath the mobile transport robot detect the magnetic field emanating from the metal strip, utilizing it as a guiding track.

Should the AGV veer off the intended path, the sensors adjust the steering motors to realign the mobile robot, ensuring it stays on course. This enables transport systems to memorize routes and replicate them in subsequent operations. The magnetic track maintains consistent polarity: the upper side is magnetized northward, while the lower side is southward.

During AGV system implementation, companies need not adhere to any specific floor requirements; the magnetic track can be applied to any surface. However, in industrially utilized halls, steel reinforcements are often present to enhance the load-bearing capacity of concrete components.

In some instances, steel reinforcements may become magnetized, potentially creating interference fields that adversely affect the AGV’s magnetic sensors. Although such interference is typically minimal and does not impede the AGV’s functionality, mobile transport robots are equipped to handle such scenarios optimally. Most AGVs come equipped with multiple localization methods by default, enabling seamless navigation even in areas with potential interference from steel reinforcements.

Close-up of magnetic track (Image: © SAFELOG).

Optical localization

For optical lane tracking localization in mobile robot systems, cameras play a crucial role. Two primary configurations exist: front-facing cameras directed along the AGV’s path of travel, or downward-facing cameras positioned to capture ground images.

With a forward-facing camera setup, the AGV promptly detects and approaches individual carts, often utilizing markers like ArUco for identification. This configuration enhances operational flexibility, especially for brief deviations from predefined paths. Optical localization with a front camera is particularly advantageous for quick off-track maneuvers.

Previously, workers meticulously marked the AGV’s destination, but now, rough planning suffices. Once the AGV reaches a certain distance from the cart (usually 1-5 meters), aided by the front camera, it swiftly locates the cart and moves in for pickup.

Conversely, downward-facing cameras enable highly precise AGV positioning and localization, commonly using GRID localization. This method involves placing magnetic points or barcodes at one-meter intervals on the floor, allowing AGVs to align and determine their exact location. Optical localization can also use floor markings recognizable through signal colors or barcode tapes. A distinct contrast between the floor covering and the marking ensures effective track detection by sensors.

Downward-facing cameras are especially beneficial in warehouses with multiple shelving rows. The GRID system allows AGVs to accurately pinpoint their position within each row, facilitating smooth navigation along extensive aisles.

Laser localization

In this localization method, a laser scanner is affixed to the AGV. The scanner initially maps the operational environment, identifying its spatial contours. The AGV software then stores this data, distributing it to all other AGVs operating within the same vicinity, thereby ensuring consistency across vehicles.

Laser localization serves a dual purpose: firstly, it facilitates the creation of a map essential for setting up the mobile robot system. Secondly, it ensures seamless operation by comparing the laser-detected environment contours with the stored map, achieving precise AGV localization in space.

Contour-based laser localization proves especially advantageous in areas with stable contours, where fixed reference points such as pillars suffice for effective utilization. The AGV’s broad laser field of view minimizes interference, even when encountering passing vehicles.

However, challenges arise in dynamic environments with frequent object movements or variable contours, such as areas with high pallet activity. In such scenarios, the AGV’s accurate positioning may be compromised if objects within its field of view obstruct its sightlines, leading to localization issues.

In general, the effectiveness of contour-based laser localization improves with the presence of numerous fixed points in the operating environment of the mobile transport robot, ensuring smoother navigation.

Triangulation

Triangulation works similarly to GPS, with the AGV calculating its position by measuring distances from three spatial points.

Triangular reflector as an enhanced safety measure for improved detection by mobile robots (Image: © SAFELOG).

What navigation options are available?

Effective localization is crucial for precise navigation, determining the AGV’s route to its destination. Factors like speed, efficiency, or route length guide the navigation system’s path calculations.

How do autonomous and track-guided navigation differ from each other?

An AGV system offers the choice between autonomous navigation or following a designated track. The functionalities vary as follows:

Autonomously navigating vehicles have the flexibility to move freely within an area without being confined to predetermined tracks. However, this mode is seldom utilized in logistics settings due to its process-oriented nature.

In warehouses and production facilities, strict regulations govern traffic flow involving AGVs, manually operated transport vehicles, and personnel. To ensure precise coordination among all participants, companies opt for track-guided navigation for AGV operations.

Track-guided navigation resembles a highway system, with vehicles traveling from a starting point to an endpoint along predetermined lanes. Similarly, in production environments employing AGVs, track-guided navigation enables them to choose the quickest and most efficient route for their journey.

What is the difference between a virtual and a physical guideline?

There are two methods for conducting track-guided navigation: utilizing either a physical guideline or a virtual one to navigate through space.

In the classic use case, an employee lays down a magnetic track on the warehouse floor before the AGV system is activated. The magnetic track is detected by a corresponding sensor in the AGV, allowing the mobile robot to commence operation promptly. Up to this point, the process is physical in nature. The vehicle identifies the physical guideline on the floor and adheres to the specified paths.

Alternatively, when a virtual guideline is preferred, the AGV traces the outline of the magnetic track and converts it into digital data. The traveled routes are digitally recorded and constitute a map that the AGV can reference for orientation.

Distinction is made between permanent and temporary virtual tracks. Permanent virtual tracks are permanently stored in the digital map. Using a map editing tool on a laptop, additional permanent tracks can be incorporated into the digital map, which can be adjusted as required.

Moreover, temporary virtual tracks are employed for singular tasks that necessitate deviation from the usual route. Once the AGV completes the task, the virtual track stored in the map is removed. For instance, digital tracks can be created leading to a QR code attached to a transported cart. The AGV follows this track to the destination, and upon reaching the cart, the temporary virtual track is deleted.

Absolute and relative track guidance

Absolute lane tracking precisely determines the driving path along the guideline, guiding the AGV to its destination point.

Relative localization allows the AGV to deviate from the predefined lane. For instance, if the AGV is autonomously approaching a cart located outside the traditional path, this is managed through relative lane tracking. Barcode markers are affixed to the cart and detected by camera systems, signaling the AGV to depart from the guideline. The AGV then travels to the cart using the spatial marker and picks it up.

Combined localization and navigation methods at Safelog

At SAFELOG, our driverless transport systems leverage a blend of magnetic and laser localization techniques to ensure seamless operation. This combination proves especially effective in environments prone to physical wear or interference. With this hybrid approach, our AGVs navigate reliably under diverse environmental conditions, providing uninterrupted service.

Warning message from the mobile robot upon loss of track detection (Image: © SAFELOG).

Summary

The choice of localization method and whether to use physical or virtual guideline navigation depends on the specific automation project and the sector’s particular requirements. When planning, it’s essential to develop the best concept tailored to the individual needs of the customers.