Developing AMRs and AGVs for Warehouses and Manufacturing Facilities

• Off-the-shelf components for integrating vehicle buses and sensors like cameras, lidar, or laser scanners
• Easy integration of specific I/O to enable navigation required by automated guided vehicles
• Comprehensive localization and path detection algorithms like SLAM for developing navigation algorithms for automated mobile robots
• Develop your solutions faster using a graphical block-based framework tailored for multisensor applications

Task

Future warehouses (Logistics 4.0) and factories (Industry 4.0) will be automated to a high extent. Automated mobile robots (AMRs) and guided vehicles (AGVs) will be integrated into this concept to move goods from one place to another, even in hostile environments and rugged conditions. Whether they are guided or not, for safety reasons these vehicles will be equipped with the sensors required to autonomously detect persons or obstacles and stop in case of emergency. 

Challenge

AGVs and AMRs are similar in some respects. Both have to be equipped with sensors like cameras or ultrasonic sensors that allow them to detect obstacles along their path and stop or move around them if required. They also need bus networks, such as CAN, that are used in the automotive domain. A key difference is in the way the vehicles select and follow the path between their starting point and destination. The navigation software of automated guided vehicles is straightforward and requires additional infrastructure on site to work properly. AGVs use specific onboard sensors for navigation. The sensors detect markers like transponders, reflective tapes on objects, and magnetic or colored tapes placed along their predefined path. In contrast, automated mobile robots use more complex and rugged path planning and navigation that are based on algorithms like SLAM (simultaneous localization and mapping) to allow for highly flexible trajectory calculation with no need for any supporting infrastructure in the warehouse or factory. To satisfy the requirements of perception-and SLAM-based navigation, AMRs are often equipped with more advanced sensors, such as stereo cameras, lidars, or laser scanners. 

Solution

The multi-modal sensor development and execution framework RTMaps enables easy integration of vehicle sensors, I/O, and vehicle buses such as CAN that are necessary to develop applications for AGVs and AMRs. RTMaps by design manages asynchronous data streams and provides them in a time-correlated manner as input for perception, data fusion, and trajectory planning algorithms. In just a few clicks and with minimum configuration effort, you can easily drag the required components from a commercial off-the-shelf (COTS) library, add them to an application diagram, and connect them to your algorithms. For the development of path-planning and navigation functions, the RTMaps AI Store provides dedicated algorithms from our partners for SLAM, drivable surface detection, and many more use cases. This means you can directly develop and execute your complete application in RTMaps on the dSPACE automotive-grade prototyping platforms MicroAutoBox Embedded PC or AUTERA with the click of a button. The platforms offer a number of interfaces to connect vehicle buses and sensors of different types. Their GPU and computation power can be tailored to individual requirements. The comprehensive library included in RTMaps provides further components like data viewers, recorders, or players, and allows for a connection to simulation environments. This enables not only prototyping, but also data logging and replay so you can test your applications for AGVs and AMRs in just one tool.