Aleksandar Dovenski (2012) Implementation of pointing at objects with a robotic arm. EngD thesis.
Abstract
The subject of the gradation thesis is pointing at an object with the robotic arm Katana. The robot autonomously detects the object in its environment, using a camera and points at it with his end effector. The system was developed in Robot Operating System, a software framework for developing robotics software. The robotic arm, alongside its environment, was simulated using Gazebo, a three-dimensional multi-robot simulator. The objects on the table were detected with the usage of the depth sensor Kinect, which provides depth data in the form of a point cloud. After the object is detected, the searches for an inverse kinematics solution to move the end effector towards the object, pointing at it. After such a solution is found, a motion planning algorithm for moving the robotic arm from its current state, to the desired one is run. Lastly, Gazebo plays the joints' trajectories, thus simulating the robotic arm's movements.
Actions (login required)