Team 8013 is the Boston Lions. We have significantly grown our team this year and have not had much time to maintain a build thread. In the spirit of better late than never here are some updates from our season. We are also excited to compete in the Springfield District Championship this season for our first time ever. We wanted to share some of the things we’ve come across this season.
In our first 2 seasons we ran tank drive systems after the 2022 season our team was excited to move to a swerve drive solution. We could see the time pressure during the competition season was a tough time to develop a new skill so we took time over the summer to learn how to build and run a swerve drive system. Based on teams we talked with we decided to use West Coast Products Swerve X system.
The team studying the swerve drive system from over the summer:
The other off-season project we tackled was building a proper cart for our robot. In particular we wanted one with some storage, and a tray for our control system. It needed to be light and able to be transported in a car. Here’s a photo of the build:
There are two new machines we have this year. One is a ShopBot CNC router which cuts aluminum very well:
We also got a MarkForge printer that can print onyx which we have used extensively:
The very start of the season had us building various game elements that we knew we would later need for practice. We were very pleased that this year we did not need to find practice space with 20’ high ceilings (last year was extremely hard to find practice space for). Within the first week we had the main game elements built.
The season started with a lot of debate around an intake system. We mocked up a few ideas and pushed further on ones we thought had potential. The shape an orientation of cones was always a major headache. This is a “push around” robot that we mocked up to see if a similar intake from our 2022 robot might work, we were also curious of the shape of the “hopper” we take game pieces into made a difference in helping to orient cones:
We really liked the idea of an intake that could hand cones/cubes off to an arm that then placed them onto the scoring spots. Here’s another drawing a student did:
The other main topic of debate was the arm. We had seen teams in 2022 have success climbing with the West Coast Products telescoping arm, so we purchased one of those kits and decided we would base our design around that geometry. We did spend a few days looking at telescoping elevators teams had designed in the past that are driven both in and out, but after a few days could see that the complexity of designing a system like that for us was going to be very difficult.
So with the spring driven telescoping arm as our base we started playing with different geometries. The team was keen on designing a low center of gravity robot so we played a lot trying to figure out how to have an arm that with a pivot that was low to the robot. At the time we were also quite keen on an intake system and making the arm and intake geometry work together was difficult. After several days of iteration a student drew this as an option that we ran with:
After a week or more playing with 4-bar linkage intake systems, hoppers, grippers and arm geometry with had this early prototype done in CAD that we were happy enough with to start building:
As our building began we were very focused on saving weight. We spent a lot of time machining metal away from parts. In particular we heavily machined our arm. Long term this turned out to be a waste of time. While it saved weight that durability trade-off was not worth it and we had enough other unknown variables that we should have been focusing on. Long story shorty we need to remember to get something built that works well first and then worry about saving weight vs. focus on it too much early.
Our jig for machining various arm elements. You can see here an example of us cutting away excess metal:
We ran some tests on different cutters. Our favorite aluminum cutter is the 1/8" single flute carbide bit you can buy on Thrifty Robot. We run this bit between 20-30 inches/min and .02 depth of cut, we tested some 3 flute and others and they just don’t work as well. Image below of speed tests from 15-35 inches/min and .01 to .02 depth of cut.
We’ve realized through this season that standardizing on a small set of screw types helps a ton with simplicity. At this point we restrict ourselves to 8-32, 10-32 and 1/4"-20 screws. Here’s an early build version:
We continued with our build, at this point we were still very keen on an intake. You can see the intake we designed on this version:
At this point we had a robot that could drive around and we could start to test with. We had a few big observations:
- The arm was very shaky and when extended did not hold position well
- Our gripper as designed was not working (more on that later)
- Our intake system didn’t work great and the torque on the shaft that lifted it up and down was too much (the shaft had stripped). We ultimately ended up scrapping the idea of an intake. There are still relics of it if you take a close look at our frame today.
As we mentioned we focused too much on weight savings to the detriment of durability. The torque on our arm had bent the main supports:
We wanted better leverage on arm movement which resulted in us creating a large cog which is a fairly unique feature of our robot this year. As a side note years ago we use to make crepes before doing robotics meetings. We would joke that we should be building a crepe maker for a robot. So we decided to name our original robot “the crepe maker”. As we have gone through subsequent seasons references to crepes are sort of a side joke on our team, so when we were designing a really big gear the similarity with a large crepe wasn’t lost on us. Hence the name.
The large gear worked! We cut it on our mill with a 1/8" bit and sanded the sides slightly with an angle grinder, it has 227 teeth and is ~18" in diameter. We run #25 chain.
The other major arm upgrade we did was adding a slip-clutch to the main drive gear so that under significant loading the arm could move without causing things to bend. Had we added this first the light weight design we had done likely would have worked fine, without a slip-clutch the excessive load on the arm was a major issue. We used this slip clutch from McMasterCarr. We did have to create a custom shaft that could go from 1/2" hex to 3/8" spline shaft. We were able to get this manufactured on https://www.xometry.com/ for ~$85 a shaft.
The installed slip clutch looks like this. We have it connected to a stack of the REV Robotics Planetary Gear system that is 3:1, 3:1, 3:1 gears, that is attached to a MAX 90 Degree Gearbox which is then attached to a the 22 tooth gear that comes with the slip clutch. The chain off this turns our 227 tooth large main gear. So off the motor shaft we have something like a 300:1 gear ration which has decent speed and good strength.
Our gripper has been its own saga. One of our original goals was to design our robot with no pneumatics. The prior 2 years we used pneumatics a lot and were curious to see if we could do without this year. We found some nice looking linear motors on AndyMark that looked like they might be swap in replacements for pneumatics. So we designed around this and built a gripped that looked like this:
Initially this seemed to work as expected. One of our students even figured out how to watch for a voltage spike off the power supply module to know when it was gripping. This worked quite well. As we ran with this design though we started to find that the life span of our linear motors was terrible. There was a warning on AndyMark that these motors didn’t like to be held in a “stalled” state which we found we needed to do in order to hold sufficient grip on a cone. We tried a variety of changes to the motor control hoping to get around this, but in the end decided it was just not that durable solution we needed to we worked on a plan B.
Plan B was to try NEO 550 motors as the drive mechanism. We were using a Neo 550 as the pivot motor on the 1st design so we added a 2nd that would open and close the gripper. This design looked like this (we actually built all of the ones below unfortunately we didn’t take photos of the builds):
We could see that the torque NEO 550s put out was significant and were excited that this design might work quite well. We did not appreciate that NEO 550s also did not work well in a “stalled state” while they didn’t break they just didn’t generate much holding torque. We were trying to operate the NEO 550s as if they were a stepper motor, we were trying to get them to just turn a few turns at a time in a controlled manner. This design was a great learning experience for how NEO 550s and the Spark Max Controllers work, but it wasn’t work-able as a gripper solution. So we went looking for a plan “C”.
Our 3rd attempt was to replace the NEO 550s with a geared motor. Through the kit of parts and First Choice over the years we had accumulated a few of the Johnson Electric PLG Gearmotors. So we created a design that let us swap these in for the NEO 550s. It looked like this:
This worked a lot better than the NEO 550 solution. The pivot motor worked great we started by using the encoders that are built into it. These were OK, but we ultimately changed to using a CANCoder absolute magnetic encoder, our swerve drive system used these so we were familiar. These helped because we could understand pivot position when the robot started.
Unfortunately the gripper solution in this design still did not work well. So we started looking for a plan “D”
Our fourth design swapped the spur-gear design for a worm gear design that looked like this:
This version also beefed up the gearing that the paddles are attached to as well as swapped the metal paddle arms for polycarbonate paddles. These turned out to be a fortunate change because as were later learned they were the weakest component in the design, so when we rammed into something these paddles were the first thing to break, this has helped protect the remaining components from damage to some extent. We were pleased that we had found a work-able design. Someone on our team remarked that with these paddles our robots gripper looked a bit like a duck-bill maybe we should think about naming it the “Boston Duck Bot”
We were hoping to post as the season went along, but hopefully the above is helpful to anyone interested. We completed our 2nd competition in Revere MA last weekend. We’ve made numerous improvements and fixes as the season has gone along but have been quite pleased with how things are holding up so far. Here’s a photo of the robot and the team from the last competition:
Very interesting concept and thanks for sharing. I see many teams adopt an all BLDC motor solution and to me many times you need to just find the right motor for the application. This simplicity of a brushed motor and the fact it has a thermal switch makes it a good solution for the lower power applications. I’ve seen many of these high power motors (775, 550 neo, etc) go up in smoke so that’s where a motor with a build in thermal switch is a nice feature.
It’s nice to see so much variety in the solutions this year. Seems to be one of the benefits of a pick and place game.
Great job at districts! Super fun design that made you guys stand out and we hope to see you again next year.
- 8046 the lakerbots