Robot fleet control for mining applications using OpenRMF and OpenTCS

In a mining environment, reducing as most as possible the human contribution for task execution is a key aspect, since most of the operations could be life threatening. The purpose of this thesis work was the development of an application devoted to the control of multi-fleet robotic systems operating in a block caving mine, accounting for the execution of tasks of different type and the traffic issue. Two applications were developed, respectively relying on the OpenRMF and OpenTCS frameworks, based on different scheduling algorithms, to determine which of the two helps achieving the best improvement in production, quantified in terms of total cycle time. This was made possible after generating a simulated environment using Gazebo 11, obtaining a scanned representation of the mine using a 2D LiDAR sensor, and creating a topological map to give special properties to the locations in which tasks are intended to be performed. Finally, the navigation procedure was accomplished exploiting Nav2 navigation stack, accounting for a self-localization process. Eight different simulations were performed to test the applications performances, each one repeated multiple times to assess the results repeatability. The systems were tested employing an increasing number of machines, and generating conflicts among the paths the machines have to follow of variable complexity. From the simulations, OpenTCS performances resulted to be unaffected by the number of machines, but unable to solve all the possible types of conflict. On the other hand, OpenRMF resulted to be able to solve all the possible conflicts but not optimally, yield less predictable results and to degrade its performances as the number of machines increases, till leading to a crash when this number gets larger or equal to 6. Both the framework presented advantages and limitations for fleet control, making OpenRMF best suited when the number of machines is low and there are complex conflicts, and OpenTCS when a higher number of entities is necessary.