Autonomous mobile robots for human-aware intralogistics and inspection operations

Mobile robotics is a research domain that focuses on developing autonomous vehicles that can navigate purposefully over large unstructured spaces to develop a task. These devices require perception, localisation and mapping, path planning, and motion control. Real-world robots also need to interact with people. To do so, they would benefit from having human-robot interaction capabilities. Nonetheless, current autonomous mobile robots lack many of the required elements for this. The paragraphs below summarise how each section of this thesis has addressed a research gap to provide solutions to these issues. Robust person detection and tracking is a complex task that can be crucial in many real-world applications. Generally, autonomous mobile robots detect people using depth cameras, lidars, or a combination of both. The first chapter introduces a novel omnidirectional system to detect people in all directions. To improve people localisation, a multi-modal algorithm using 2D lidar data has also been developed. Finally, this person detection system has been used to guide an autonomous mobile robot following a person. During experimentation the standalone omnidirectional system has proved very effective at detecting people, even though 2D lidar data has been found to be necessary for robust distance estimation. Overall, the results demonstrate encouraging person detection capabilities. Material handling, or intralogistics, are all activities related to the movement of materials in a factory. Autonomous mobile robots enable the automation of such non-value adding activities, even in crowded manufacturing facilities. Ideally, these robots would move considering the comfort of the workers. Human-aware navigation algorithms have been previously presented to enable this. However, current approaches do not address many common industrial tasks. This section utilises the omnidirectional system to detect people in hazardous areas, such as docking stations or working areas, and behave accordingly. During testing the presented approaches managed to effectively react to the presence of people in industrial environments and avoid dangerous situations. Finally, autonomous mobile robots have been used to inspect an assembly task in real time. Currently, aerospace structures are manually equipped by trained operators. To improve current methods, an automated system that ensures quality control is proposed. An autonomous mobile robot follows operators and localises the position of the inspection using a thermal camera and 2D lidars. Then, an in-process monitoring algorithm uses depth camera 3D measurements to check if the system has been correctly installed and provide feedback to the workers. During testing, the systems have proven to be reliable. Overall, the results show that the solution is promising for industrial applications.