To hasten the process of saving lives after disasters in urban areas, autonomous robots are being looked to for providing mapping, hazard identification and casualty location. These robots need to maximise time in the field without having to recharge and without reducing productivity. This project aims to improve autonomous robot navigation through allowing comparison of algorithms with various weightings, in conjunction with the ability to vary physical parameters of the robot and other factors such as error thresholds/limits. The lack of a priori terrain data in disaster sites, means that robots have to dynamically create a representation of the terrain from received sensor range-data in order to path plan. To reduce the resources used, the affect of input data on the terrain model is analysed such that some points may be culled. The issues of identifying hazards within these models are considered with respect to the effect on safe navigation. A modular open-source platform has been created which allows the automated running of experimental trials in conjunction with the implementation and use of other input types, node networks, or algorithms. Varying the terrains, obstacles, initial positions and goals, which a virtual robot is tasked with navigating means that the design, and hence performance, are not tailored to individual situations. Additionally, this demonstrates the variability of scenarios possible. This combination of features allows one to identify the effects of different design decisions, while the use of a game-like graphical interface allows users to readily view and comprehend the scenarios the robot encounters and the paths produced to traverse these environments. The initially planned focus of experimentation lay in testing different algorithms and various weightings, however this was expanded to include different implementations and factors of the input collection, terrain modelling and robot movement. Across a variety of terrain scenarios, the resultant paths and status upon trial completion were analysed and displayed to allow observations to be made. It was found that the path planning algorithms are of less import than initially believed, with other facets of the robotic system having equally significant roles in producing quality paths through a hazardous environment. For fixed view robots, like the choice used in this simulator, it was found that there were issues of incompatibility with A* based algorithms, as the algorithm’s expected knowledge of the areas in all directions regardless of present orientation, and hence they did not perform as they are intended. It is suggested that the behaviour of such algorithms be modified if they are to be used with fixed view systems, in order to gather sufficient data from the surroundings to operate correctly and find paths in difficult terrains. A simulation tool such as this, enables the process of design and testing to be completed with greater ease, and if one can restrain the number of parameters varied, then also with more haste. These benefits will make this simulation tool a valuable addition to the field of USAR research.

Since their introduction in the 1950s, industrial Automated Guided Vehicles (AGV) have gone from automatic machinery limited by hardware to complex robots limited by software, called Autonomous Mobile Robots. Small and medium businesses need to be able to utilize cutting-edge technology. Therefore, this research focuses on deploying mapless AMRs on cheap open source AMRs in dynamic industrial environments. The study implements Dijkstra’s and A-STAR algorithms on a simulated Turtlebot3 model deployed in a Gazebo rendering of an industrial warehouse with moving objects added. The Turtlebot3 model traverses the environment where time and distance results are observed. The results shown in the research indicate that Dijkstra’s algorithm is barely affected by the change of the initial map state, while the A-STAR algorithm performed worse on average. Future work should focus on minimizing the sensors needed and continue testing with more algorithms, but early tests show promising results.