Please use this identifier to cite or link to this item:
http://cimat.repositorioinstitucional.mx/jspui/handle/1008/1034
Optimal Sampling-based Motion Planning with Applications to Nonholonomic Dynamical Systems Under Visibility Constraints and Object Reconstruction Optimal Sampling-based Motion Planning with Applications to Nonholonomic Dynamical Systems Under Visibility Constraints and Object Reconstruction | |
HEIKEL YERVILLA HERRERA | |
Acceso Abierto | |
Atribución-NoComercial | |
CIENCIAS DE LA COMPUTACIÓN | |
In this work, we study properties of local planners for nonholonomic dynamical systems to achieve asymptotic global optimality through a sampling-based planner, namely, the RRT*. A concise summary of the conditions found in the literature for the RRT* to asymptotically converge is presented, along with further formal results about local planners such as the finding that the topological property is not a necessary condition in the current context. Besides, an experimental set-up is used to evaluate the performance of the local planners in terms of resulting trajectory cost and time of convergence. Furthermore, a complex robotic system is contemplated, namely, a mobile manipulator robot that includes visibility constraints in the planning requirements. Relevant concepts for complex robotic systems such as detachability and time dominance are introduced. Experiments in a physical mobile manipulator robot, equipped with a camera in the arm's end-effector, are shown, which validate the proposed theoretical modeling. Also, two aspects of motion planning for object reconstruction are investigated. First, the effect of using a samplingbased optimal planning technique to move an 8 degrees of freedom mobile manipulator is studied in terms of several performance criteria. Based on those criteria, the results of the reconstruction task using rapidly exploring random tree (RRT) approaches are compared, more specifically, RRT* smart versus RRT* versus standard RRT. Based on our results, it is concluded that the use of the RRT* approaches improves the measured performance criteria compared with a standard RRT. Second, the problem of defining a convenient stopping probabilistic test to terminate the reconstruction process is addressed. We propose two probabilistic stopping tests for the 3D reconstruction process. Comparisons between both termination criteria are done and the results are analyzed. The simulation experiments show that the two proposed stopping tests are adequate. They stop the reconstruction process when all the portions of the object that are possible to be seen have been covered with the field of view of the sensor. In addition, in this work, we propose a new objective function to be optimized, which requires to find the time optimal trajectories in the presence of obstacles for a nonholonomic robotic base satisfying visibility constraints at the sensing locations. One of the termination tests together with a whole method for object reconstruction is expe | |
15-11-2019 | |
Tesis de doctorado | |
OTRAS | |
Versión aceptada | |
acceptedVersion - Versión aceptada | |
Appears in Collections: | Tesis del CIMAT |
Upload archives
File | Description | Size | Format | |
---|---|---|---|---|
TE 759.pdf | 13.04 MB | Adobe PDF | View/Open |