Mobile Robot Autonomous Environment Perception And Navigation In Indoor Environments

This dissertation presents recent research on the problem of environmental modeling for exploration,localization and map building for a fully autonomous and self-contained mobile robot.The robots acting in an indoor building environment.They have a multi-sensor setup where encoders are used to extract the odometry data and a front positioned 2D Light Detection and Ranging(LiDAR)sensor are employed to perceive the surrounding.Intended to study mobile robot exploration,localization and mapping,the work begins with the research around techniques to register the environment with laser scans besides the dynamically and statically noise presence,followed by the propose of different architecture systems allowing two types of mapping to be merged to find the localization of the robot.This system presents the possibility of creating small map data files to register the environment.The low data density favors the efficiency enhancement during data processing.The representation of the environment around is critical for mobile robot navigation.The task defines which perception capabilities are needed and which navigation technique is permitted to use.The present environmental modeling register is both in metric and topological formats.By coherently combining the two well-known technics,advantages of both approaches are added to confront the drawbacks of a single methodology.The capability of the hybrid system is exploited to model an indoor building environment where the metric information is used locally inside the “Rolling Dynamical Window”.This format is constrained to the robot surroundings while the topology of the whole environment is accurate separately,resulting in avoiding the need for global metric consistency.Metric maps register just enough information to create the topological nodes and edges in a relational map.This allows the mobile robot to use the neighbor's topological relations in navigation,which reduces the resulting path after the route planning.This hybrid model permits the robot to start the routine from zero,without the requirement of previous information or mapping.For this,it uses the integrated and standard software,ROS linked with modules to explore the environment,to map it,and safely move through it to achieve the desired destination based on tasks for autonomous navigation.The theory is hypothetically proven,however,as stated during the work,the robot needs to justify the theory eventually in the practical applications.This extensive experimentation performed with fully autonomous self-contained robots and the experiments were subsequently carefully analyzed.The topological and metrical approaches were successfully tested in both simulated environments and real-world applications,which indicated the effectiveness of the proposed method.