IOC Robotic's Lab | Projects / Motion Planning using PMD's

HomePage WikiSandbox Lab Robots Haptics Cameras Hands BMM 1 & 2 Drones Ontology KnowledgeMotionPlanning RFID Pant-tilt Pozyx Development tools ARM development STM32F4-Discovery CMake ? Revision Control ROS Orocos ? Soem Realtime OS Fril ? Emacs with Lisp Projects Cgrip ? Platform ? Visual Odometry Dual-Arm Motion Planning PhanTorque Libraries 3DOF Regressor Bi hand-arm PCA Kinect 3D Model Capture Tips ? Console LaTeX Working on ? Project Grasping edit SideBar	*Motion planning of dual-arm robot using Principal motion directions.* This work deals with the problem of simplifying the motion planning of anthropomorphic robot with two arms. The proposed solution reduces the dimension of the search space to simplify the problem. This reduction is performed using the concept of Principal Motion Directions (PMD), that are obtained applying the Principal Component Analysis (PCA) to samples of robot movements, obtained from real movements of two human arms while doing a specific task. Using this reduction, a solution to the problem of motion planning is obtained in a more efficient way. To sum up, the steps included in this approach are: the capture of the work space for two human arms doing a task by means of trackers; reduction of the planning space dimension by means of the use of principal motion directions; and planning of the movement of two anthropomorphic robotic arms using a sampling-based roadmap planner considering the principal motion directions. In this work, we have implemented a tool to obtain the Principal motion directons associated to a determinated task. This is done through the following steps: 1. Capture of the movements of a person performing a determinated task in a real work space. This is done using two trackers, the tracker Fastrak of Polhemus and the tracker Flock of Birds of Ascension. 2. Compute the value for each joint of a model of a robot like there are doing the same task done for a person in the real work space. To implement the model of the robot, we use the Kinematics and Dinamics Library (KDL). ROS are used to connect with the nodes of every tracker, and obtain the information relative to the pose of the trackers. 3. Apply the principal component analisis with the information of the joints doing a determinated task to obtain the principal motion directions associated to this task.

Motion planning of dual-arm robot using Principal motion directions.

This work deals with the problem of simplifying the motion planning of anthropomorphic robot with two arms. The proposed solution reduces the dimension of the search space to simplify the problem. This reduction is performed using the concept of Principal Motion Directions (PMD), that are obtained applying the Principal Component Analysis (PCA) to samples of robot movements, obtained from real movements of two human arms while doing a specific task. Using this reduction, a solution to the problem of motion planning is obtained in a more efficient way. To sum up, the steps included in this approach are: the capture of the work space for two human arms doing a task by means of trackers; reduction of the planning space dimension by means of the use of principal motion directions; and planning of the movement of two anthropomorphic robotic arms using a sampling-based roadmap planner considering the principal motion directions.

In this work, we have implemented a tool to obtain the Principal motion directons associated to a determinated task. This is done through the following steps:

1. Capture of the movements of a person performing a determinated task in a real work space. This is done using two trackers, the tracker Fastrak of Polhemus and the tracker Flock of Birds of Ascension.

2. Compute the value for each joint of a model of a robot like there are doing the same task done for a person in the real work space. To implement the model of the robot, we use the Kinematics and Dinamics Library (KDL). ROS are used to connect with the nodes of every tracker, and obtain the information relative to the pose of the trackers.

3. Apply the principal component analisis with the information of the joints doing a determinated task to obtain the principal motion directions associated to this task.