The project started off with the creation of several objectives. These objectives have shaped the contents of the project in several major ways which will be described further below. but first a general review of the workflow during the project.
The objective to create a torque resistant material:
With the connex pinter we could make use of printing more materials in one print. Using this fact the main focus of the tests was to make the finger move more human by making it torque resistant. After some trial and error the added shape of a double helix in the design seemed to be the most effective of all the tests.
The objective to create a more humanoid feeling:
Because of the fact that the project had to be based of the pneumatic technique that was also used in the original project it was rather hard to create a humanoid feeling. specially with the added boundaries the connex printer added with material properties restrictions. because of all those problems the creation of a humanoid form was scrapped in favor of the functionability of the gripper.
Multi-material printing has a great potential to improve soft robotics in functionality and interaction. Due to the composition of the materials, functional components can be create in one print with unique material characteristics. Although it is warranted that one should be careful with the use of the materials as for example the materials from the Object500 Connex are very brittle and can break quite easily. The removal of support material can be quite an issue, however water soluble support material is in attendance and should ease the development of soft robotics with multi-material printing and enable the designer to print even more complex geometries.
An important part of a pneumatic system are valves. These were examined to test which is the most air-tight. Depending on the manner of connection (tapping or screwing) a softer or harder material should be chosen. Further, printing circular forms perpendicular to the printing direction, will cause oval-like shapes in soft and flexible materials due to the weight on top of the print. This phenomenon could be observed in a prototype of a John-Guest connection.
Below you can see the cross-sections of the different connections, from top to bottom post threaded connection (used in the one-way valves and gripper), pre threaded connection and a John Guest connection.
Post-Tapped Thread Connection
Pre-Tapped Thread Connection
John Guest Connection
We explored if it is possible to print one-way valves. Two prototypes were printed and both works surprisingly well. Although some air is still leaking out (because of the cleaning process to remove support material as it damaged the enclosed parts), it demonstrates that functional parts can be printed quite easily.
Below you can see the cross-section of both valve designs. The blue parts are the enclosed and can move around. The reds parts are seals made from softer material so the moving part can fully close one direction if pressure is applied from one side.
Ball One-Way Valve
One-way valve based on a free-moving ball in an enclosed space. The ball is not able to close the upper output due to the form of the enclosed space, however it is possible for the ball to close the bottom output because of the sealer.
Plug One-Way Valve
One-way valve based on a plug-like object. The plug can make an up an down movement to change the state of the valve. The top part is completely solid (and can therefore close the valve), while the bottom one has air inlets.
To get an idea of how our models will behave once printed, a Finite Element Model Analysis (FEA) was conducted with the gripper-finger models. This analysis is a non-linear regression model, which can simulate quite accurately large and complex deformations in materials, under complex load applications and boundary conditions. As we do not know the right material properties just yet, we have applied representative figures for the properties. This is to get an idea on how the bellow shapes will behave in reality (i.e. to know if we can expect a bending motion or a stretching etc.)
Right now, one simulation was done to test how well the simulations work. This simulation can be watched below. It should be noticed that the program does not take self inflicted collisions into account yet. This means that the solid can intersect itself when blown up, which would in reality induce more deformations in bending direction for this model.
Screenshot of a FEA in CATIA
Animated FEA in CATIA
Today, the first model of a gripper-finger was made in Rhinoceros and sent to the Objet500 Connex printer (1). The model is visible in the figure below. This printed model is used to test the initial behavior of the material and air chambers.
The model of the gripper-finger was printed later that day. However, while removing the support material one bellow broke immediately, making it already unable to function. Thereby we had issues to completely remove the support material on the inside of the finger in the air-chambers.
Therefore the following was concluded from this first prototype:
- The thickness of the finger is to small. It broke too quickly. So for the next model a thickness of 2 mm will be applied (instead of 1mm).
- Also the finger was to flexible. It could not stand straight on its own. For a next prototype we will have to make the material less flexible. That means the material for the next prototype will be a combination of 70% ridged and 30% soft material, instead of 60%/40%.
- Lastly, we need to think of a way to successfully remove the support material inside the finger. First ideas are to add a ridged cylinder in the whole length of the finger which can be pulled out after printing. This will ensure that there is space to erode support material. Secondly, a one way valve at the top of the finger could be added, so water can flow through the whole finger and clean all the air chambers, but prevent the air from leaking when in use.
For the next prototype these findings will be used to improve the gripper-finger.
Rhino model of the printed finger.
The print from the Rhino model. Although it looked good on first sight, there it is more difficult to make a working finger than we thought. For example one bellow broke immediately when it was cleaned (broken bellow is shown in the last two pictures).
(1) Stratasys, http://www.stratasys.com/3d-printers/design-series/connex-systems/
Today we conducted a tensile strength test on the three specimens which we made previously (1). We started our testing at the faculty of Industrial Design Engineering (IDE) with a universal testing machine (UTM) which was made by a student. However due to problems with the software, we could not get results with it. We therefore went to the 3me faculty to do the test with a ZwickZ100 UTM (2) in combination with an extensometer (which was used to more accurately measure the change in length of the specimen).
From this test we learned that the ridged specimen breaks at 40.64 MPa and the composite at 10.27 MPa (however the last one broke not in the middle, but the bottom part was ripped off). The UTM machine was not able to measure the most flexible material as no force was build up for measurement.
Next week, we will do more tensile testing, again with the soft material, but also with another new specimen (60% ridged, 40% soft material) as this will probably be the composite material we will use for the gripper-finger.
First tensile strength test set-up at IO. However because of software problems this did not work well.
The ZwickZ100 UTM just broke the rigid specimen in half.
Remains of the specimen.
(1) Weblog post, 1 October 2015, https://softrobotics2015.weblog.tudelft.nl/2015/10/01/connex-printing-test/
(2) Zwick Roell Group, http://www.zwick.com/
Today we did a test with the Objet500 Connex 3D printer to see how the printing process works and what the effect is of using rigid and flexible materials.
We printed three specimen (rigid, flexible and a composite) to use it for a tensile strength test next week. The pictures below show the results.
The most interesting print was the composite as with the structure inside it you can easily twist the specimen, but its inside structure prevents bending. This is done by making rigid cylinders (cores) over the length of the print and bind them together with more flexible material. The cylinders gives the composite the property to resist bending (as it is rigid over the length) while the flexible material can make the cylinders move dependent from each other and therefore twist.
Our specimen prints are processed by the connex printer
The printing result. The pink material is support material which needs to be cleaned with a water jet ot by scratching it off. The total printing time was 30 minutes.
This picture shows how the composite specimen can easily be twisted due to the combinations of rigid and flexible materials. Bending it is quite impossible though.