| Resumo: | Human-robot collaboration is a field with multiple applications, and this varies in complexity from the medical field to entertainment. In the entertainment field we can achieve human-robot collaboration by playing games, which presents enough complexity for a first contact with the subject for someone who is starting to dive into the subject. Example of that is the project developed by IRIS: a robotic arm in collaboration with a human solves a pentomino puzzle. That project inspired this thesis by trying to achieve a wider application than just one collaborative game and apply it to different hardware. This thesis achieves human-robot cooperation by playing different board games, allowing a collaborative yet competitive solution. To do so, I used Kinova's Jaco assistive robotic arm and a Kinect camera sensor. I divided the developed work in three major topics: manipulation, perception, and gameplay. To manipulate the Kinova's Jaco arm, I used its ROS package in the ROS framework. With the Kinect sensor and with the use of PCL and OpenCV libraries, I was able to locate the pieces and board used for each game, allowing the robot to manipulate without help from the human. For the gameplay I explored AI algorithms and implemented them to allow the robot to plan its own moves. Using the ROS framework allowed me to connect the three different parts of this project.
|
|---|