| Περίληψη: | The purpose of the present work is to design and build an integrated robotic vehicle navigation system, as well as to construct two-dimensional and three-dimensional indoor maps.
The robotic vehicle should know where it is at all times on the map so that it can compute a route to go from point A to point B, while avoiding any new obstacles on it’s way that are not on the maps.
It should also be able to solve the problem of “kidnapping the robot” that is, moving the robot to a new position, the robot can re-localize it self, three-dimensional features are very helpful in this area.
The robot is a 4-wheel-drive differential (meaning that the wheels don;t turn) so it’s dynamic model is relatively simple.
The robot is also equipped with quadrature encoders on each wheel to measure the rotation of each wheel, thereby computing “blind” odometry and giving us the capability to do a close loop pid control to keep the robot’s velocity constant.
It has a 2-d lidar to measure the distance everywhere around it on a plane, as well as a 3-d RGB-D camera (kinect) to create the 3D maps.
We will introduce simultaneous localization and mapping or SLAM and Path Planning techniques, so that the robot will be able to map and autonomously navigate unknown areas
Furthermore, we will see how the robot can recognize 3D objects and their relative position in 3D space. This will be achieve with SVM classifiers which will use features of hsv-color and shape-normal in a histogram form for training and classification.
Lastly we will attach to the robot, a robotic hand of 6 degrees of freedom so that it can interact with it’s environment transforming it into a mobile manipulator.
The brain is the Jetson Nano embedded computer with 4 cores at 1.3 Ghz and 125 cores gpu providing us with the computing power we need.
We will do all of this by taking advantage of the Robot Operating System (ROS) capabilities that help us quickly program robots with packages designed by others, as well as a system to communicating the various operations between them.
Reading this thesis will familiarize you with the capabilities of ROS, and you will be able to build similar robots yourself in a methodical way, without even having previous experience in robot design and programming.
|
|---|
