Autonomous Mobile Robots (AMRs) can locate and navigate themselves in an environment—this is how they accomplish their tasks of assembly and transportation. But how do they know where they are and where they need to be? Behind these simple questions lies the science of localisation, mapping, and navigation. This blog post explains how our AMRs, the arculees, locate, map, and navigate to ensure accurate, efficient, and autonomous movement.

Autonomous Mobile Robots (AMRs), like the arculees, must navigate different environments, such as warehouses or other facilities, to move from one place to another and perform a task. The first step to navigation is mapping, which simply refers to the process of creating a map. Before arculees can drive autonomously, we need a map because otherwise, we don't have a point of reference; we don’t know where the robot is and where it needs to go.

How do arculees Generate a Map?

Mobile robots rely on data from their surroundings to create accurate maps for navigation. They need two types of information: their own location and the location of other objects around them. The arculees collect this important sensor data through two LiDAR (Light Detection and Ranging) scanners, an IMU (inertial measurement unit), and the wheel odometry.

LiDAR Scanners: They use laser beams to measure distances to objects, generating a point cloud that is used to create a map using the robot’s software.

Inertial Measurement Unit (IMU): The IMU is an electronic device that provides data on a robot’s acceleration and rotation.

Wheel Odometry: It is the process of using data from wheel encoders to estimate the robot's position and orientation over time by calculating how far the wheels have travelled, typically relative to the robot's starting point.

Together, these tools help the arculees create a map representative of their real environment and determine their position within it.

SLAM - Simultaneous Localisation and Mapping

The method of collecting data from scanners, IMU, and wheel odometry to generate a map is called SLAM. It stands for simultaneous localisation and mapping. To create an accurate map, you also need to know where the robot is on that map, which is what localisation stands for in the acronym. With the scanners and sensors in place, you drive the robot around and collect the scanner data and the data on where the robot is. Finally, you combine that to create a map.

Once the map is there, the robot software loads the map as a reference frame and is then able to drive within it. Since the map itself has a coordinate system, the software can tell the robot where it should move and where it is, creating a route and driving from A to B.

Tools and Techniques: Mapping, Localisation, and Navigation

Localisation, mapping, and navigation are the processes that go hand in hand to ensure autonomous driving of mobile robots like the arculees. However, there are certain tools and techniques to make the processes work:

At arculus, our tools, algorithms and sensors include the SLAM toolbox, Google’s Cartographer, IMU, wheel odometry, and two laser scanners. The SLAM toolbox is an open-source 2D SLAM tool specifically developed for Robot Operating System 2 (ROS2). Both Cartographer and SLAM tools are used for real-time localising and mapping across multiple sensor configurations and platforms.

Meanwhile, there are two ways to record a map:

Offline: We create a recording by driving around the environment whose map we need and collect data. This data includes scanner information, IMU (inertial measurement unit) readings, and wheel odometry. Afterwards, we process the recording using a Cartographer to generate an offline map.

Online: We primarily use Cartographer for online recording. It simply processes live data from the robot in real time to create the map.

Once the arculees have a laser map, they can drive automatically. However, to complete tasks, operate safely, and avoid deadlocks, an efficient traffic management system overlooks the robots, giving them specific driving orders. This system lives in our fleet manager. The best path is calculated by considering various aspects such as battery charge, distance, and possible driveways where the robots can go. Therefore, our Fleet Management Software sends step-by-step instructions to the AMR, guiding it for the next actions at a time. As the robot moves, the fleet manager continuously updates the route, ensuring safe, smooth, and accurate navigation.

Current Challenges and Hopes

Like any other technological advancement, AMR mapping and navigation comes with challenges, but there are always ways to find better, more viable solutions. While our current mapping techniques work perfectly within small to medium areas, a possible caveat can be creating maps of large and dynamic environments. However, our team of developers at arculus addresses this by recording the data and later processing it offline on more powerful computers, where computational constraints are no longer a concern.

Although the present mapping processes are effective, arculus aims to improve and innovate continuously. That is why our ultimate goal is to keep updating and refining the AMR mapping methods.

Hopes for the Future

Future advancements in mapping and navigation promise faster algorithms and higher map accuracy. Improvements in accuracy can address persistent issues. For example, enhanced precision would enable robots to navigate more efficiently, reducing errors and the need for manual corrections.

These advancements hold great potential for the role of autonomous mobile robots. With greater accuracy, robots could move faster and operate even more reliably. As a result, accuracy can create a more stable and efficient solution for dynamic environments.

The Way Forward

Mapping and navigation in autonomously driven vehicles is an emerging but crucial part of robotics. As such, it has ample room for growth and advancement. While there is no silver bullet to improve everything all at once, the tools developers use for mapping and navigation are composed of smaller components and algorithms, so a slight improvement in one of the parts can positively affect the entire process. On the other hand, staying connected to recent and relevant research, reading publications and keeping oneself up-to-date on what’s happening in the field is essential for future breakthroughs. There is much room for improvement, and tools need updating, especially with artificial intelligence entering the equation.