Please use this identifier to cite or link to this item:
http://hdl.handle.net/11375/25997
Full metadata record
DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Bone, Gary M. | - |
dc.contributor.author | Guo, Peige | - |
dc.date.accessioned | 2020-10-23T19:19:10Z | - |
dc.date.available | 2020-10-23T19:19:10Z | - |
dc.date.issued | 2020 | - |
dc.identifier.uri | http://hdl.handle.net/11375/25997 | - |
dc.description.abstract | To fully exploit the advantages of human-robot collaboration the robot must be allowed to move when the human is close to it or even in contact with it. This thesis presents the development of a collision avoidance system which addresses the safety problem for the case of one person sharing a workspace with a robot manipulator. The system consists of a depth camera that measures both the colour and depth of the scene near to the robot, a laptop computer and several software algorithms. A human modeling algorithm generates a plane model and union of spheres model from the point cloud. Sphere-swept lines are used to geometrically model each link of the robot. Their position and orientation in space are calculated using the robot’s joint position measurements and its kinematic model. Two collision avoidance algorithms are presented for controlling the robot’s trajectory based on the geometric models for the human and robot, and the robot’s desired task. The first collision avoidance algorithm solves the inverse kinematics problem and avoids collisions using an expanded version of the manipulator Jacobian matrix. A second collision avoidance algorithm using nonlinear model predictive control is developed as an alternative approach. The algorithms have been implemented in a simulated environment which includes a human working in the shared workspace with a simple planar robot and with an Elfin 5 industrial robot. A variety of scenarios are simulated and the results are compared. The simulation results showed that the first collision avoidance algorithm may be computed fast enough to be applied in real-time and worked well for static or slowly moving obstacles. The second collision avoidance algorithm had superior performance when the obstacle was moving and when the simulated robot had a realistic time delay. However, its computation time was too long to be used in real-time. | en_US |
dc.language.iso | en | en_US |
dc.title | Collision Avoidance System for Human-Robot Collaboration | en_US |
dc.type | Thesis | en_US |
dc.contributor.department | Mechanical Engineering | en_US |
dc.description.degreetype | Thesis | en_US |
dc.description.degree | Master of Applied Science (MASc) | en_US |
dc.description.layabstract | Robots are expected to become a significant part of our society in the future. They will need to interact and collaborate with people at home and at work. This will require them to rapidly adapt to dynamic situations. This thesis is concerned with solving this problem for a robot arm working close to a person. A depth camera measures colour and depth of the area around the robot. A software algorithm is presented to model the person using this colour and depth information. Two software algorithms are presented to control the robot arm. They try to move the end of the robot arm towards a target location while simultaneously avoiding it colliding with the person. They are tested using a simulated industrial robot arm. The results show that the algorithms working together can prevent collisions between the person and robot, while simultaneously moving the robot towards its target location. | en_US |
Appears in Collections: | Open Access Dissertations and Theses |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
Peige_Guo_thesis_2020_MASc.pdf | 4 MB | Adobe PDF | View/Open |
Items in MacSphere are protected by copyright, with all rights reserved, unless otherwise indicated.