Automated Manufacturing of Cementitious Vessels

3 Degree of Freedom Robotic Arm Made from Scratch for STAR Project

PROJECT ONGOING

PROJECT PURPOSE
I have been a member of the concrete canoe team since October 2013, and have been an active member since then. Drexel University's Concrete Canoe Team (CCT) is one of the only teams in the world to stray away from the more primitive, hand-placing techniques for fabricating canoes, but rather uses High Volume, Low Pressure (HVLP) sprayers in order to apply the concrete to a male mold. More information regarding CCT can be found below in the section entitled 'Concrete Canoe Team'.

A member of the CCT team in Spring of 2014 asked me how difficult it would be to automate this system. I performed research on this, and decided to make the automated manufacturing of cementitious vessels the project topic for a selective summer research program, called STAR, for freshman. During STAR, I learned about the groundwork needed to launch a project like this, as well as lean about kinematics, basic computer vision, and robotic arms. I plan on taking this further and in the future apply for funding to purchase an ABB industrial robot with a track in order to be able to implement this system into Drexel's concrete canoe fabrication process.


PROJECT PROGRESS

STAR RESEARCH
During the STAR program, I learned about what would be needed in order to created an automated system for fabricating concrete canoes. I determined that this system would be rather easy to implement since the CCT already uses a spraying system, which can be automated without the need of changing the process that is already in place.

I determined that this system would require two 6 Degree's Of Freedom (DOF) robotic arms that move along a track, giving the robot a 7th DOF. Using 7 DOF avoids any problems that would be caused by gimbal lock. There are 6 equations that determine the location and orientation of every location within 3-space, which will always have solution if 6 DOF were to be used (not including cases where gimbal lock occurs). The reason for using two of these is that our current system of fabrication requires two sprayers for an optimized product since the quality of the canoe is highly dependent on the drying time given for each layer (more information can be seen in the section below entitled 'Concrete Canoe Team'). If the spraying process for a given layer were to take too much time, the canoe could not be optimized. This is the reason for using two 6 DOF robotic arms. Though, if the spraying process could be done within the required time using only one 6 DOF, there wouldn't be a need for the second one.

Another important component to the system is quality control. There need to be a robust way in order to determine the thickness of application for each layer, and be able to determine which areas need to be sprayed more. This can be done manually, but this would then need to be fixed manually since this information would be difficult to try to communicate to the robot. Automating this system is an option, but then would require computer vision. This can be done using a Kinect camera, which is able to give distance data, color data, and IR data. The distance data can be used to determine the thickness of the current layer and output a graphical representation of the data in order to let the CCT know which area need to be fixed. If this process were to be automated further, the robot could be able to apply more concrete in the areas that are not thick enough.

The final major component of this automated system is the concrete mixing and spraying processes. The mixing process must be manual due to the amount of R&D needed to create an robust automated system. Currently, each mix is checked and fine-tuned by the mixing manager to make sure each mix is as identical as possible as the preceding mixes. This can be manually done, but would require for the spraying system to have some sort of tank that material can be added into. The spraying system would have a constant nozzle pressure if a gravity fed system were to be used, and the ability to change the nozzle pressure if a pumping system were to be implemented. The latter would require a system that would not clog since around 60% of concrete mixes consist of aggregates. A entire system for quality control for the spraying process would be needed, which I was not able to find. The former constant-pressure system was determined to be most efficient due to this case, unless a pumping system described above already exists on the market.

In order for this system to be implemented, research would needed to be conducted for the computer vision system for quality control, a spraying system, and a software that could integrate both of these as well as perform the calculations for the kinematics, path finding, and determine the state of each part of the system. This software would determine the location of the male mold in 3-space, determine the path needed to spray the entire canoe for each layer, spray the canoe and determine any areas that need more material, wait for a layer of reinforcement to be placed by the CCT team, and repeat until all layers have been placed.

In terms of deliverables, forward and inverse kinematics were developed in MATLAB for a 3, 6, and 7 DOF system, basic coding for computer vision and distance graphs using a Kinect sensor was created, and a 3 DOF physical model was fabricated using 3D printed parts and dynamixel motors. A demonstration showing the range of motion of the physical model can be seen here. The abstracts, paper, poster, and timeline for this project can be seen below in the uploaded files section. More information regarding the STAR program can be seen below in the section titled, 'STAR PROGRAM'.

If you would like to see the MATLAB code developed for the robotic arm kinematics before I clean up the code to make it more user friendly, feel free to contact me and ask for them. You can contact me using and of the links I have put on the home page.

ABB Industrial Robotics
While on an internship at University West PTC in Trollhättan, Sweden from September 2014 to March 2015, I had the opportunity to be trained on working with ABB industrial robots as well as robot simulation and programming within ABB's RobotStudio. This has given me the understanding of the capabilities of industrial robots, and has shown me that many operations determined important during the STAR research program has been simplified with this program. All kinematics, path finding, working with sensors, and working with models in 3-space can be performed in RobotStudio. This program also has many libraries for spraying systems that needs simple modifications in order to work with other systems. All operations described in software needs of STAR research project can be performed by this program.

From here, simulations can be made to show that this is possible and will be used in order to try to secure funding. The type of robot as well as a layout of the spraying system will need to be determined before moving forward. This will require some calculations on regarding force produced by HVLP sprayers, as well as reachability needed in order to be able to spray all parts of the male mold.


MORE INFORMATION
Concrete Canoe Team
Drexel University's Concrete Canoe Team (CCT) performs research on concrete mix design and fabricates a concrete canoe every year. Mix design research includes finding different types of cementitious mixes, aggregates, and accelerants that will produce a lightweight, yet strong canoe that is able to maneuver through water. Since 2013, Drexel's CCT has moved away from a hand-placed fabrication style to using High Volume, Low Pressure (HVLP) sprayers for a shotcrete application. Switching to this method allows for a more homogenous canoe, which is able to be fabricated in a fraction of the time in comparison to the fabrication of a hand-placed canoe. The shotcrete application does require a different style of mix since it must be able to be shot from a spray gun, meaning that large sized aggregates cannot be used and the mix must be less viscous.

The Drexel CCT competes in the National Concrete Canoe Competition (NCCC) each year, placing 1st at regionals, and 1st in the design paper, 10th overall at nationals in the spring and summer of 2014. More information regarding the NCCC can be found here.

STAR PROGRAM
I took part in a selective freshman summer research program where accepted students find a professor within the university who is willing to be a research advisor, and decided on a research project together with their advisor. During this program, the Office of Undergraduate Research (OUR) who created the STAR program, holds different workshops in order to develop research skills of the students within the program. At the end of the whole program, there is a day of poster presentations where students print a poster and talk to others at this event about what they have accomplished during the summer. This is followed by advisor of the year awards and a closing ceremony.

LINKS

NOTES
I will upload MATLAB code for Inverse and Forward Kinematics developed for the robotic arm.
ĉ
Frederick Wachter,
Nov 2, 2014, 6:35 AM
ĉ
Frederick Wachter,
Nov 2, 2014, 6:31 AM
ą
Frederick Wachter,
Nov 2, 2014, 6:45 AM
ĉ
Frederick Wachter,
Nov 2, 2014, 6:31 AM
Ĉ
Frederick Wachter,
Nov 2, 2014, 6:48 AM
Comments