Automated Fiber Placement
Touch based object shape perception and grasping
Automation of Composite Manufacturing Processes for Aerospace Structures Using a Parallel Robot
This project aims to address automation of composites manufacturing processes for advanced aerospace structures based on robotic technologies, which involves multi-disciplinary research topics including materials, robotics, aerospace and manufacturing. Compared to current hand lay-up techniques, automation in composites manufacturing is vital to improve quality, repeatability and reduce costs. A comprehensive characterization study of novel fibrous materials will be conducted to investigate the effects of material variability on composites processing. Based on material models, we will investigate and explore mechanical properties of Parallel Robots in Automated Fiber Placement (AFP) process by enhancing the current techniques that use serial robots to deposit fibers onto a mold. Although serial robots have simple structure and control, they generally have low stiffness and large inertia, due to this optimum compaction and precision cannot be guaranteed in AFP. In contrast to this, Parallel Robots have multiple support limbs with high stiffness, good accuracy and high speeds. The characterization studies based on novel materials will benefit the design and optimization of the new Parallel Robot for AFP. This work will provide a unique demonstrator with huge potential of applications in local industry, and will pave the way for future ground breaking research and commercialization of composites manufacturing.
Collaborations: Aerospace Research Innovation Center, Khalifa University
Project Fund: Khalifah University Internal Research Fund (KUIRF) Level 1
People : Dr. Dongming Gan, Dr. Rehan Umer, Prof. Lakmal Seneviratne, Prof. Jorge Dias
- D. M. Gan, J. S. Dai, Jorge Dias, R. Umer, and L. D. Seneviratne, “Singularity-Free Workspace Aimed Optimal Design of a 2T2R Parallel Mechanism for Automated Fiber Placement”, Transactions of the ASME: Journal of Mechanisms and Robotics, 2015, in press.
- D. M. Gan, J. S. Dai, Jorge Dias, and L. D. Seneviratne, “Forward Kinematics Solution Distribution and Analytic Singularity-Free Workspace of Linear-Actuated Symmetrical Spherical Parallel Manipulators”, Transactions of the ASME: Journal of Mechanisms and Robotics, 7(4), 2015, pp. 041007_1-8.
- Y. F. Zhuang, D. M. Gan, “Unified Singularity Modeling and Reconfiguration of 3rTPS Metamorphic Parallel Mechanisms with Parallel Constraint Screws”, Advances in Mechanical Engineering, 2015, 7: 352797.
- D. M. Gan, J. S. Dai, Jorge Dias, and L. D. Seneviratne, “Constraint-Plane-Based Synthesis and Topology Variation of A Class of Metamorphic Parallel Mechanisms”, Journal of Mechanical Science and Technology, 28(10), 2014, pp. 4179-4191.
- D. M. Gan, J. S. Dai, Jorge Dias, and L. D. Seneviratne, “Unified Kinematics and Singularity Analysis of A Metamorphic Parallel Mechanism with Bifurcated Motion”, Transactions of the ASME: Journal of Mechanisms and Robotics, 5(3), 2013, pp. 041104_1-11.
Touch based Object Identification and Grasp Selection
Grasping different objects with unknown structure and material type is a challenging problem in robotics. The correct manipulation of an object by a robot depends on many factors including the object’s mass, weight, position, and rigidity among many other factors. The main challenge is to provide the robot with the ability to identify and sense different objects and grasp them appropriately. In general grasping mechanisms can vary widely depending on the consistency and nature of the object. For example, solid objects need to be grasped differently than deformable objects. Applying a touch based analysis on different objects is a technique that has been recently utilized to allow robotic systems to manipulate and interact with objects using suitable grasping mechanisms.
The proposed solution includes three different components; object segmentation, object identification/classification and grasping trajectory selection. The objective is to perform a touch based analysis of objects with different level of rigidity and determine the suitable grasping mechanism. The touch based analysis is based on fusing sensory data from vision sensor (RGBD Camera), tactile sensor and force-torque sensor. During the touch based analysis, the robotic arm motion is monitored and data is collected from the tactile sensor installed at the fingertips of the robotic hand and the force-torque sensor mounted on the wrist of a robotic arm. Recorded data is then analyzed online to classify the objects, determine its nature, and if possible identify it based on their characteristics and an objects database. After the classification/identification process, the grasping mechanism will be determined according to the object’s characteristics and shape. Figure 1 summarizes the proposed solution. Figure 2 illustrates the touch based analysis on a paper surface simulating a deformable object. The following steps are executed while performing the touch based analysis and grasping:
- Surface Segmentation
- Object Clustering
- Touch Point selection
- Recording Tactile Data
- Sensor fusion and Object Classification & Identification
- Grasping Pose Selection
This ongoing research aims at advancing current grasping and physical manipulation technologies to a state where robots will be capable of efficiently and accurately interact with objects of any type, shape or form.
People: Dr. Tarek Taha, Dr. Dongming Gan, Randa Almadhoun
- H. Hussein, T. Caldeira, D. M. Gan, J. Dias, L. D. Seneviratne, "Object Shape Perception in Blind Robot Grasping Using a Wrist Force/Torque Sensor". In proceedings of 2013 IEEE International Conference on Electronics, Circuits, and Systems (ICECS 2013), December 8-11 2013, Abu Dhabi, UAE.