Me 565 Final Project Report Shoulder and Bicep of InMoov Robot

Group Members:

Renjie Xie

Zhaolun Song

Dongyang Zhang

Jiannan Zhao

Chao LIu

Project Introduction:

Out project is about the fabrication of the biceps and shoulders of InMoov robot on right and left.

InMoov is an open source, humanoid robot, constructed out of 3D printable plastic body components, and controlled by Arduino microcontrollers. It can be separated into several assemblies. The biceps and shoulders we fabricated are parts of those assemblies

Here are the pictures shows the subassemblies of shoulder and bicep

In our project, we tried to optimize the assemblies to be more suitable for AM which is easier to assemble and fabricate. And we assumed our designs should also be fit in real production condition.

Fabrication Plan:

Thought process

In order to optimize the design, we tried to use some snap fit features to replace those screws and combine subassemblies as many as we can. But some screws cannot be removed which is very important to keep the structure stable and strong enough. And some parts cannot be combined together due to the printing orientation. As we have leant from the class that the xy direction can give the part enough strength if it is loaded part.

CAD modeling/modifications

For the shoulder shown in the picture, the screws used to connect the PivGear and PivMit endure the main shearing stress to rotate the biceps. So we decide to keep these four screws. While we have another perspective about those four screws combining the PivTit and Pivcenter together. We think the main load of these screws is in the direction along the height of the gear. And the force of whole shoulder in that direction can be distributed to other parts like the existed snap fit features and screws. So we decide to remove these screws and use solidworks to design four corresponding snap fit features for them.

For the bicep, since the higharmside and low armside, reinforcer are unnecessary separated, we decide to combine them together, in order to reduce the printing time. And for the subassembly between Rotcenter and RotTit which is similar to the PivTit and Pivcenter of shoulder, we also decide to replace these screws to snap fit for the similar reason.

We also assembled all parts for the shoulder and bicep together to verify our design and for the further assembly with other parts

Machine and material selection

Because the color of the shoulder is mostly black and shoulder is the assembly enduring most of load to rotate the left and right arms, and there is not very big plane part. We decided to use makebot to produce most of the parts of shoulder. The material offered by makerbot is PLA which is strong enough and relatively cheaper. While we would use uprint to print servoholder, because that a little flexibility will be needed to assemble two servoholders,servoholster and PivTit which can be offered by the ABS plus.

When it comes to bicep, since there are several big flat parts need to be printed, we choose dimension with ABS plus to print these parts like the armside, in order to avoid the unpredictable shrinkage. And other parts of bicep we also choose to use makebot with PLA to print, due to its relatively low cost and strength.

Initial estimates for material consumption and time

We used the makebot software with low density setting to estimate the material consumption and time

For example:

Total material estimation is 1568.6g, while the density of PLA is 20.5g/in^3, so the estimate material is 76.5 in^3;Total time estimation is 68h 10min

Fabrication process:

Orinetation and layout

Gears and worms are printed vertially. Other parts are printed horizontally.

For example:

The actual machine used, print time and material consumption

The details for every part are shown below

Priority Part Name Quantity Printer Print Date Model

Material Support material

Printing time

1 leftPivcenterV1 1 MakerBot Replicator 11/20/2015 59.54g 4h2m

2 leftPivTitV1 1 MakerBot Replicator 11/23/2015 30.53g 1h55m

3 PivcenterV1 1 MakerBot Replicator 11/24/2015 59.5g 4h1m

4 PivTitV1 1 Dimention 12/9/2015 1.62 in3 0.44in3 2h11m

5 leftPivMitV1 1 Dimention 12/9/2015 2.33 in3 0.6in3 3h27m

6 PivMitV1 1 MakerBot Replicator 11/24/2015 40.44g 3h27m

7 PivConnectorV1 (*4) 4 MakerBot Replicator 11/24/2015 77.12g 1h27m

8 PivGearV4 (*2) 2 MakerBot Replicator 11/23/2015 58.4g 3h2m

9 PivWormV2 (*2) 2 MakerBot Replicator 11/23/2015 23g 6g 1h9m

10 servoholderV1 (*2) 2 uPrint 11/23/2015 2.36 in3 0.42in3 2h1m

11 servoHolsterV1 (*2) 2 MakerBot Replicator 11/23/2015 47.24g 2h41m

12 PistonbaseantiV2 2 dimension 12/8/2015 1.31 in3 0.43in3 3h15m

13 gearpotentioV1_(1) 2 dimension 12/8/2015 0.11 in3 0.03in3 18m

14 leftRotTitV2_(modified) 1 dimension 12/8/2015 2.04 in3 0.89in3 2h58m

15 RotTitV2 1 dimension 12/8/2015 2.04 in3 0.40in3 1h57m

16 spacerV1 2 dimension 12/8/2015 0.42 in3 0.13in3 24m

17 GearHolderV1 1 dimension 12/8/2015 0.20 in3 0.04in3 33m

18 RotPotentioV2 1 dimension 12/8/2015 0.17 in3 0.05in3 13m

19 RotcenterV2 1 dimension 12/8/2015 4.85 in3 0.55in3 4h57m

20 reinforcerV1 2 dimension 12/8/2015 0.67 in3 0.13in3 37m

21 elbowshaftgearV1 1 dimension 12/8/2015 1.26 in3 0.23in3 1h48m

22 elbowshaftgearV1 1 dimension 12/10/2015 1.26 in3 0.23in3 1h48m

23 servoholderV1 1 dimension 12/9/2015 1.63 in3 0.19in3 2h21m

24 servoholderV2 1 dimension 12/9/2015 1.63 in3 0.19in3 2h21m

25 RotGearV4 1 1 dimension 12/9/2015 2.49 in3 0.8in3 5h

26 RotGearV4 2 1 dimension 12/9/2015 2.49 in3 0.8in3 5h

27 PistonanticlockV2 1 dimension 12/9/2015 1.87 in3 1.16in3 9h21m

28 servobaseV1 1 dimension 12/9/2015 1.8 in3 0.27in3 2h32m

29 servobaseV1 1 dimension 12/9/2015 1.8 in3 0.27in3 2h32m

Material usage:

PLA: 767.77g (37.45in3)

ABS plus: 48.61in3

In total: 86.06in3

9.56in3 material usage than estimated

Total Print time: 90h41m

22h31m more than estimated

Post processing

Due to rough surface finish of PLA, we spends lots of time to use the file and sand paper to file the surface of gears and worms with help of plastic cleaner to achieve better surface quality. For the support material of those parts printed with dimension, we removed what we can removed by hands and put those parts with supports which we cannot manual removed into solution. This solution took hours to solve the support. The part printed with makerbot can be manually removed with tools. The part printed with uprint can be easily removed by hand from the build tray.

30 RotWormV5 1 dimension 12/9/2015 0.79 in3 0.45 in3 3h6m

31 RotWormV5 1 dimension 12/9/2015 0.79 in3 0.45 in3 3h6m

32 lowarmsideV1 1 dimension 12/9/2015 1.99 in3 0.31 in3 1h32m

33 lowarmsideV1 1 dimension 12/9/2015 1.99 in3 0.31 in3 1h32m

34 lowarmsideV1 1 dimension 12/10/2015 1.99 in3 0.31 in3 1h32m

35 lowarmsideV1 1 dimension 12/10/2015 1.99 in3 0.31 in3 1h32m

36 HighArmSideV2 1 dimension 12/9/2015 2.47 in3 0.42 in3 1h43m

37 HighArmSideV2 1 dimension 12/9/2015 2.47 in3 0.42 in3 1h43m

38 HighArmSideV2 1 dimension 12/10/2015 2.47 in3 0.42 in3 1h43m

39 HighArmSideV2 1 dimension 12/10/2015 2.47 in3 0.42 in3 1h43m

40 RotMIt V2 1 dimension 12/9/2015 2.2 in3 0.47 in3 3h9m

41 RotMIt V2 1 dimension 12/9/2015 2.2 in3 0.47 in3 3h9m

Discussion and Conclusion: Lesson learned

1. Using simplify cad software, we can easily add the support to the part. This software can let user decide the way and the amount the support will be added. Simplify can also control the printer print the part in same work with diverse density.

2. When the build tray is not full of parts, putting parts close to the initial location of extruder can reduce the print time due to the shorter travel distance of extruder.

3. Too many big flat parts shouldn`t be put in same work in FDM AM systems. Because this will lead to high thermal stress, resulting in shrinkage of parts.

4. The printer with open printing environment is vulnerable to the change of circumstance comparing to those printer with closed printing environment.

5. Better surface quality resulting from the higher resolution of printer will makes it easier to assemble.

6. The direction of layers would affect the cooperation of parts.

Problem encountered during fabrication process

1. When printing piston base, we found that the printing orientation affect the rotation of worm:

Former chosen orientation Optimized orientation

The way worm and piston base cooperate

The reason why the worm cannot rotate smoothly is that the direction of the layer of worm and the direction of the layer of piston base are mutually perpendicular. It will add lots of frictional force.

2. The issue of modification: the snap fit features of the bicep were dissolved in the solution and the snap fit features of the shoulder broke when we tried to assemble the shoulder due to the printing orientation. We used the rest snap fit features to locate the location of screw and drill the holes for screw with tools.

3. Shrinkage issues: when using makerbot to print Pivmit, part of the Pivmit were put

outside of the paper of the build tray. This leads to less than enough adhesion to resist the thermal stress. When the lower layer cooled down the higher layer has a higher temperature. And the plastic tends to contract in low temperature and extend in high temperature. As a result the temperature distribution resulted in shrinkage

4. The problem with original STL file: The worm and gear we download from the InMoov

website are not same part used in the assembly tuition. Besides the STLfiles have some problem. The bottom teeth of the worms would have over interference with the teeth face of gear. It will stop the gears and worms from rotating. The method we took to fix the problem was using files and sand papers to file the bottom teeth of worms, resulting in slope like section teeth. And we optimized the STL file to avoid such problem in the future.

The original STL file The optimized STL file

The worm after post processing

Conclusion

During the fabrication process, we encountered any unexpected issues like the problem of original STL file. We barely found the solution of this problem two days before the presentation. It takes us lots of time and effort. Other than that, such project helped us to learn more about the way the printer works and the part quality printed by different printer. Makerbot cannot offer the surface quality we expected due to the rigid characteristic of PLA and the thicker layer diameter. In addition, it takes us lots of time for post processing. And the exposed building environment makes it more sensitive to the change of temperature. If we would build these part again, we wouldn`t choose makerbot to print the worm and gear. Also, the snap fit feature we designed didn`t end well, next time, we should use better optimized design to replace the screws. The direction of the layer is a very import factor we should take more serious next time, which affects the assembly more than we thought.