3. Manufacturing and Prototype Iterations

Construction of the First Prototype:

Once we settled on the four bar linkage design, we began construction of a simple prototype. The rigid body holding the shovel was made of two linkages with a 3D printed shovel designed to be attached to the end. So although the mechanism was functionally a 4-bar linkage, we made 6 linkages for the initial prototype. For each prototype, we modeled the whole assembly as much as possible in CAD to get a better understanding of how the components would stack up and where they might interfere with each other.

Pictured Above: The prototype design in CAD, made of six linkages including the ground and a shovel head. The support structure was not yet designed.

The linkages (including the ground linkage that held it all together) were laser cut out of acrylic and the bearings were then inserted into them. For all of the bearings, we tried our best to have them press-fit into the linkages. However, due to the finnicky and sometimes inconsistent tolerances of the laser cutter as a result of factors such as the software trace feature and the laser diameter, the fit wasn't always perfect. Even with the same hole diameter for every cut, some of the bearing holes came out with a perfect press fit while others were too loose. Whenever the holes were slightly too loose, we would add a small dot of hot glue to the hole before inserting the bearing. This would force it to stay in place. In addition to the linkages, we laser cut a few circular spacers to add to the linkages' stack where they were necessary to prevent the linkages from crossing or colliding while moving

We encountered an emergency complication for our first prototype, when we discovered the night before it was due that the steel shaft we were planning on using for the joints was made of hardened steel and would be impossible to cut on any of the equipment we had access too. With no other option, we resorted to using hot glue sticks for the first prototype's shafts.

Because we had yet to finalize what the support structure would look like, we attached the initial prototype to a cardboard box full of dirt using dowels and hot glue. This let us test if the digging movement would work on actual dirt.

Overall, the first prototype was of pretty low construction quality due to the limited time it was constructed in before the deadline and the issue we encountered with the shaft last minute. However, it allowed us to test the digging motion and acted as a proof of concept for the main linkage design.

Pictured Above: A bearing on the shaft that we later found out was uncuttable (left), The first version of the Shovel head attached to the end of one of the laser-cut acrylic linkages (right).

Pictured Above: The final assembly of the first prototype. The shafts are made of hot glue sticks, and the assembly is attached to a cardboard box with hot glue and dowels. Though the prototype was crude in its construction, once the box was filled with dirt it acted as a useful proof-of-concept of the core digging mechanism.

Further Prototyping Until the Final Product:

Rebuilding the Core Mechanism:

We took the feedback we received on our initial prototype into consideration as we started on the next build. This next prototype evolved and changed over time as new parts were added to it and modified until it eventually became our final product.

Pictured Above: New linkage designs. The linkages were reworked into just four main links, one of which was uniquely shaped to hold the shovel in the right position and orientation. The ground linkage was reworked to be longer. The grounded points remained the same, but the part itself was made bigger so that the shovel had more room to move. Supports for either side were also designed to hold the mechanism at the desired height.

Trying to avoid the issue we had with the uncuttable steel shaft, we ordered new 6mm ID bearings and new aluminum 6mm OD shaft. However, when they arrived in the mail, we were dismayed to find that the shafts would not fit in the bearings without the use of a vice To get around this issue and make prototyping easier, we purchased 1/8" diameter wooden dowels and sanded them down to a press fit in the bearings. While the wooden dowels were not as sturdy and reliable as metal shafts would have been, they provided us with the versatility and customizability that we desperately needed at this stage of the process because they could be easily cut and sanded down to the size we needed.

The linkages and supports for this next prototype were designed in Solidworks and laser cut out of plywood.

Plywood was chosen at this stage for is low cost and customizability. We were designing a lot of features as we went at this point, so we wanted the option to be able to screw into it, sand it, and add holes as we found it to be necessary. The tradeoff was that this material was quite rough, creating friction between joints that served to slow the mechanism down.

Like with the first prototype, we ran into the same issue with cutting the bearing holes: the hole sizes just weren't consistent when cut by the laser. Some holes would press-fit the bearings perfectly, while others had to be held in place by hot glue. Still, we made it work. The three linkages that used to work together to hold the shovel were then redesigned into one linkage, as was advised by our TA during the prototype demo. All of these redesigned resulted in a prototype for the new core mechanism made of four main linkages, with the shovel attached to the middle linkage by a rigid body.

Pictured Above: The new laser cut plywood linkages, as well as the laser cut plywood support structure. Wooden dowels were used for the shafts at the pin joints because we could not get the aluminum shaft we ordered to fit consistently.

Adding In the Seed Dropping Feature:

At this stage, we began thinking of how we would time the seed dropping and tried to incorporate it into our existing design. Seeing the way the shovel linkage rotated during it's motion inspired us to use the orientation of the mechanism to our advantage. Inspired by those little mazes that you flip around and tilt to navigate a metal ball through them, we designed and 3D printed a plastic "maze" that would fit inside the linkage that held the shovel. While at the top of it's rotation, the entry hole would be pointed upwards for the seed to be dropped inside Then, after the shovel scoops away the dirt, the linkage will rotate naturally and the seed would roll downwards out the exit hole. This would consistently drop it at the point we wanted (between the digging out and replacing of the dirt).

Pictured Above: An image of a ball maze puzzle used as inspiration (left), and the model of our 3D printed "maze" that went inside of our mechanism (right).

Pictured Above: The orientation of the maze feature when the seed is added through the entry hole (left), and the orientation of the maze feature after the linkage rotates, allowing the seed to fall out through the exit hole (right).

We unfortunately had to 3D print the maze twice, because a small error while modeling it in CAD resulted in the holes for the shaft in the feature not being aligned properly. We remade the model, double-checked that it would fit by adding it to the prototype assembly with all the proper mates, and then reprinted it. With all the linkages and the maze printed and cut, we redesigned the shovel head slightly to fit the new design. Now it and the maze would be sandwiched between two plywood laser cut layers. With the linkages and maze all designed, we assembled them all together using the bearings we ordered and the wooden dowels.

Pictured Above: A closeup of the new shovel iteration duct taped to the mechanism (left), and a picture of the second iteration of the maze being 3D printed (right).

Incorporating the Motor:

The next step was then to incorporate the motor. To do this, we needed a motor mount that could mount on to the ground linkage and hold the motor in place. We designed this in CAD, going off of the current assembly and designing it to attach to what we already had, and we then 3D printed it. We found that the motor mount worked great for holding the motor in place, however the motor was still free to rotate within the circular mount, which became a problem. To fix this, an additional part was designed that would slot onto the existing motor mount. It had holes for two tiny M2 screws that would fit into holes on the motor and prevent the motor from being able to rotate inside the mount.

Pictured Above: The motor being held by the 3D printed motor mount.

Pictured Above: Assembly of the motor, motor mount, and additional part created to prevent rotation in Solidworks. The motor was modeled based on measurements taken of the real thing with a caliper. The other two parts were then modeled based off of the motor model's geometry, which ended up being quite accurate because all of the parts fit together well on the first iteration.

With that issue sorted, we then needed a way to snugly attach the motor to the grounded input link, so that the motor could spin it. We did this with a 3D printed part that was then glued into the bearing hole on that linkage. We had a couple of iterations of this part because we found that unless the part fit very tightly on the motor, the motor would slip inside of it and it wouldn't be able to move the mechanism. We finally solved this issue by making the hole for the motor D-shaped like the motor shaft and printing it with zero tolerance (line-to-line) so that it would be press-fit. Still, even with a perfect initial fit, we still found that over time, continued use began to wear down the soft 3D printed plastic, making the fit looser and looser. So perhaps once the right fit was found, we should have switched to a less soft material than plastic.

We then assembled everything together. Many of the elements of this build had to be screwed on, but we couldn't have the screws sticking out as it would get in the way of the linkages. We fixed this by using flat head screws and countersinking them into the plywood. This was done in an admittedly very crude and "DIY" way by carving the holes out with a knife, but it worked.

Pictured Above: The backside of the robot, showing how the motor mount was screwed on The orange piece is the part created to stop the motor from rotating in the mount.

Troubleshooting the Core Design:

After assembling the entire mechanism with the motor, we found that the whole mechanism was cantilevered and leaning away from the motor due to the weight of the linkages and shovel. Unfortunately, there wasn't enough space and length on the motor to have the shafts held up by multiple points, so we couldn't solve this issue that way. The best we could do to eliminate this problem would be to have the mechanism be as close to the mount as possible so it can't lean too far. We had incorporated spacers into the design to keep the shovel far enough away from the mount that it wouldn't collide with it while digging. Eliminating these spacers would accomplish our goal of bringing the whole mechanism closer to the mount to combat the cantilevering, but we still needed away to keep the shovel from colliding. We accomplished this by redesigning the shovel head, so that while it still attached to the linkage the same, the head was offset to the side so that it wouldn't collide with anything. We also added screw holes to this new shovel head design, so that it could be mounted in a less permanent way than glue. By making these changes, we were able to reduce the cantilevering significantly.

Pictured Above: The third and final design for the shovelhead.

Working Towards Our Stretch Goal:

After the motor was added to the assembly and we fixed the issues with the motor slipping and shafts cantilevering, we started working on our stretch goal of getting the whole mechanism to move every time a cycle was completed.

Using some 3D models for bevel gears we found online, and 3D models of regular gears that we generated online and then modified to fit what we needed, we modeled a gear assembly that would be able to translate the constant input motion of the motor into intermittent rotation of the mechanism's wheels. The bevel gears were then 3D printed, and the big and small gears were laser cut. We went through a couple of iterations for the gears as we fine tuned the design and as some of the first iterations got worn out or broken. We were originally going to just 3/4 of the teeth off of the big gear by hand to achieve the desired effect, but once we figured out how to edit the gear CAD we found online, we just remade it with most of the teeth removed. Also, since we had some trouble with the small gear not staying in line with the bevel gear, we eventually just 3D printed a new part that combined the bevel gear, small gear, and shaft.

While testing, we ran into a problem with the long shaft that ran across the back of the mechanism that was made to be able to move both wheels at once. It was made out of a wooden dowel, because it was the longest thing we had that fit the bearings. However, this meant that it was able to bend, and we ran into an issue where it would bend so much that the bevel gears wouldn't stay aligned. We solved this by designing a couple of 3D printed parts that would hold the shaft in place and prevent them from bending. These parts were then screwed into the plywood ground mount similar to how the motor was. These parts ultimately worked very well to stop the shaft from bending.

Pictured Above: Different iterations of the gears used.

The belt system that was made to translate the motion of the long shaft to the wheels was made using 3D printed band gears and rubber bands. We later replaced the 3D printed band gears with spools made out of multiple laser cut concentric circles glued together because the 3D printed ones were too rough and the supports were too difficult to remove. For the wheels meant to roll the mechanism forward, we used two plastic car wheels that we already had because the tires would provide traction, and then we used two laser cut wheels of the same diameter for the other wheels.

We ran into an issue with the wheel shafts being cantilevered, which was causing problems with getting the wheels to move. We solved this by laser cutting another one of each of the prototype supports to put on either side. That way the wheel shafts were being held up in two places, and the shafts were no longer cantilevered.

Robot Mechanism Design