DISSERTATION | tesehugocosta

THE PROJECT

PROBLEM

The industry is full of fixed robotic manipulators that perform repetitive and dumb function within a limited area and almost never interacting directly with humans. This kind of configurations are particularly useful in assembly lines where the robot have a delimited work volume, but when we speak about constructions areas where robots have to be moved around and need to interact with human operators to perform and assist in tasks, there any almost no one in work in industrial environments in exploration.

This two aspects, mobility and interaction requires the implementation of safety. Mobilities implies that the robot must be able to locate himself at all time, planning its trajectory and executing it without colliding with the environment maintaining its physical integrity and most important, cannot hurt a human in the process. Interaction is more directional to the user, the robot should be able to assist the human user without hurting him in the process. In the beginning robotic manipulators were put inside a cage, the size of its working volume, to perform their task, but when someone enter this space the robot stop immediately to prevent injuring the user. Then some or part of this cages were replaced by infrared curtains but the working principal remain the same. This only allow interaction throw predetermined zones that are available to one of the interested parts, security by segregation. Later this "rigid" working volumes were evolving to non rigid allowing a more emerging interaction since the user could enter some of the robot space, but it stops automatically when the critical area is breached. This methodology allow the user to interact with the robotic manipulator in a larger area, for example a robot that follow the user and stop when is "touched". This interaction can be achieved by instrumenting the robot in order to have perception of the humans around, but this interaction doesn’t need to be physical, the robot can interact by projecting information or working in cooperation with humans. It only need to track the position of every one of them and don’t harm them.

MOTIVATION

Video projectors are an off-the-shelf and low cost solution for machine to user interface. Adding a well located robot and with knowledge of the surrounding environment allow a very promising interaction with humans.

Nowadays when a worker need information to do its job have to consult a computer or papers, but with a robot with a projector on it this task is facilitated. The robot can project the desired information on the working area. This can go from projecting data from CAD files to instructions about how to do some tasks. This system allow information to be transmitted from the robot to the user but the opposite can also be achieved equipping the robot with sensor for user data input such as a simple mouse, a camera to detect user movements or a pointer.

For example, the system can be used to project positions for stud welding, for placement of power lines, and others, it’s also possible to assist the human in repairs by showing the built-in infrastructures. In the first case, if the CAD file are not correct the user is able to change it by giving simple instruction.

This process improve work efficiency by allowing faster access to data and quicker understanding of them.

The objective of this work is the development of an user machine interface based on projection mapping to perform in industrial environments.

The system should be able to project undistorted images under any surface from the user point of view. In other words, it should be capable to project 3D virtual images on non-planar surfaces.

To achieve this goal it is needed to calibrate the projector and camera, to create a virtual representation of the real scene and apply the projection mapping algorithm to project correct images from the user point of view, also should be able to read CAD files and extract the necessary information.

In the end, this system will be integrated in the "THE CARLoS PROJECT".

OBJECTIVES AND TARGET STATE

STRATEGY

The system should be able to project any data desired by the user on a non-planar surface.

To achieve this goal, the methodology used is based on a simulated environment of the real scene that will produce the images for the projection mapping.

To do this, the project will be separated in the next steps:

Camera projector calibration
Create a simulation of the real world in gazebo
Create a ROS interface
Import cad files to easily adapt the system for different environments and data to be presented
Improve projector localization in the real environment