China
Africa
Explore chapters and articles related to this topic
Software library for path planning in complex construction environments
Published in Jan Karlshøj, Raimar Scherer, eWork and eBusiness in Architecture, Engineering and Construction, 2018
K. Kazakov, S. Morozov, V. Semenov, V. Zolotov
To identify path conflicts the motion planning theory and software tools should be applied [2]. Unfortunately, motion planning problems are PSPACE-hard. Even being formulated in local statements, these problems can cause serious computational difficulties. Popular software libraries such as Motion Planning Kit (MPK), OpenRave, Open Motion Planning Library (OMPL) are basically intended for such local formulations [3, 4, 5]. Mathematical arsenal of the libraries is mainly based on sampling and searching techniques such as RRT (Rapidly Exploring Trees) and PRM (Probabilistic Roadmaps). These demonstrate high efficiency in disparate applications such as humanoid robotics, automotive manufacturing, computational geography, computer graphics, computational biology, but fail in complex construction environments with non-trivial topology and dynamic behavior.
Warehouse Automation: An Example
Published in Laxmidhar Behera, Swagat Kumar, Prem Kumar Patchaikani, Ranjith Ravindranathan Nair, Samrat Dutta, Intelligent Control of Robotic Systems, 2020
Laxmidhar Behera, Swagat Kumar, Prem Kumar Patchaikani, Ranjith Ravindranathan Nair, Samrat Dutta
In the case of industrial manipulators where one does not have access to internal motor controllers, motion planning refers to providing suitable joint angle position (or velocity) trajectories needed for taking the robot from one pose to another. In other words, motion planning becomes a path planning problem which is about finding a way to point poses between the current pose and the desired end-effector pose. The problem of generating collision free paths for manipulators with increasingly larger number of links is quite complex and has attracted considerable interest over last couple of decades. Readers can refer to [351] for an overview of these methods. These methods could be primarily divided into two categories - local and global. Local methods start from a given initial configuration and step toward final configuration by using local information of the workspace. Artificial potential field-based methods [352] [353] [354] are one such category of methods where the search is guided along the negative gradient of artificially created vector fields. On the other hand, global methods use search algorithms over the entire workspace to find suitable paths. Some of the examples of global methods are probabilistic roadmaps (PRM) [355] [356]and cell-decomposition based C-Space methods [357] [358]. Rapidly exploring random tree (RRT) [359] is one of the most popular PRM method used for path planning. Many of these state-of-the-art algorithms are available in the form of the open motion planning library (OMPL) [289] which has been integrated into several easy-to-use software packages like Moveit! [288], Kautham [360], and OpenRave [361].
Architecture
Published in Hanky Sjafrie, Introduction to Self-Driving Vehicle Technology, 2019
The Open Motion Planning Library (OMPL) [14] is an open source library that contains implementations of RRT, PRM and many other state-of-the-art sampling-based algorithms. As it is designed as a pure collection of planning algorithms, any additional functionalities, such as the collision avoidance function or visualization, still need to be implemented or provided by the SDV middleware.
Knowledge-oriented task and motion planning for multiple mobile robots
Published in Journal of Experimental & Theoretical Artificial Intelligence, 2019
The proposed framework implementation consists of three major layers as depicted in Figure 7: ontological knowledge, task-level and motion-level layers. The knowledge layer is coded in the form of an OWL ontology as detailed in Section 4. The task-level layer embraces the Heuristic task planner which is a modified version of the FF planner implemented using the Prolog language, and the Action reasoning process whose purpose is to determine actions conditions by calling online along offline reasoning processes. The motion-level layer comprises The Kautham Project (Rosell, Pérez, Aliakbar, Muhayyuddin, & García, 2014) (https://sir.upc.edu/projects/kautham/) that enables to plan under kinodynamic and physics-based constraints. It uses the Open Motion Planning Library (OMPL) (Şucan, Moll, & Kavraki, 2012) as its core of planning algorithms, and is integrated with the Open Dynamic Engine (ODE) for the dynamic simulations. Although any kinodynamic motion planner can be selected, KPIECE (Şucan & Kavraki, 2009) has been used in the experiments because in a comparative study (Gillani, Akbari, & Rosell, 2016) it showed the highest success rate and the best time-optimal solution as compared to other state-of-the-art kinodynamic planners. This planner does not minimise the distance and therefore the paths found will not be the shortest ones.
Automation of product packaging for industrial applications
Published in International Journal of Computer Integrated Manufacturing, 2018
C. Perez-Vidal, L. Gracia, J. M. de Paco, M. Wirkus, J. M. Azorin, J. de Gea
A repository of motion planning algorithms is available in the OMPL (Open Motion Planning Library) (Sucan, Moll, and Kavraki 2012): PRM, RRT, ESTS, SBL, KPIECE, BKPIECE, LBKPIECE, LazyRRT, RRTConnect, etc. Hence, it can be selected as the most suitable planner for the task at hand and the values for its parameters can be chosen. However, these planners may generate non-smooth trajectories and, hence, smoothing is needed before sending it to the controller.