Capstone II Spring 2024

Kate Morgan Author

06/10/2024 Added

8 Plays

Description

This is an in-depth look on the Scott Campus Capstone II Spring 2024 student event.

Searchable Transcript

Search:

[00:00:00.240](upbeat music)
[00:00:06.720]So this is object detection drone.
[00:00:08.340]So what it is, we're controlling a drone
[00:00:11.520]through our boards that we designed here.
[00:00:15.450]We're going to survey a field,
[00:00:17.640]detect whatever object you want it to detect,
[00:00:19.890]but really the idea is we wanna be able
[00:00:22.890]to identify weeds with it so that way a farmer can use it
[00:00:26.820]for weed management without having to take a plant out
[00:00:30.630]and crop dust the entire field.
[00:00:32.490]So the idea is if we can identify
[00:00:34.470]where the weeds are using AI,
[00:00:36.707]then we can reduce the over spray that would be involved
[00:00:41.070]with just applying herbicide to the entire field.
[00:00:44.520]So yeah, our device, it flies over this field,
[00:00:50.177]it'll take pictures, use that AI to identify what's there
[00:00:53.550]and then we can attribute a GPS coordinate to each detection
[00:00:58.770]so that way we can go back out and do the spray later, so.
[00:01:02.670]Bigger drone and then modify kind
[00:01:04.710]of what we've already done just to add a little bit more.
[00:01:06.450]So this doesn't have enough payload capacity
[00:01:08.490]to carry herbicide to actually spray.
[00:01:10.320]So what we would need to do is get a bigger, heavier drone
[00:01:13.050]that can handle carrying something
[00:01:14.970]and then it'd just be modified
[00:01:15.930]with what we've already done a little bit to trigger
[00:01:19.013]that herbicide to be dispersed once it gets back.
[00:01:20.640]So we use the golf balls
[00:01:22.440]because of the season that we are working
[00:01:25.184]on this project in, we're working on it over the winter
[00:01:26.356]so we're not gonna have weeds that we can detect.
[00:01:28.500]However, the nice thing is with AI object detection models,
[00:01:33.960]whatever sample set you give it, is what it can detect.
[00:01:38.310]So if we wanted to refit our model for weeds,
[00:01:43.950]all that we would need to do is get a bunch of picture
[00:01:45.990]of weeds, retrain it, and it would be the same AI model
[00:01:49.980]just with a different training set.
[00:01:51.597]And it would work the same way with our system.
[00:01:54.909](upbeat music)
[00:01:56.910]So it's like a life alert
[00:01:59.280]when somebody falls down in a small room.
[00:02:01.500]It detects the person falling down
[00:02:03.240]and then it can call contact anybody for them.
[00:02:06.690]They can set up their phone number for ultra contact
[00:02:09.720]and instead of them actually pressing a button
[00:02:12.960]and say that they fallen down,
[00:02:15.060]this system is actually built on radar.
[00:02:18.750]So it actually scans the room all the time when somebody is
[00:02:22.994]in there, always check if the person has fallen down
[00:02:25.728]and if the person did fall down, it triggers
[00:02:28.020]and then ask the person, "We detected that you have fallen.
[00:02:31.470]would you like us to call for help?"
[00:02:33.017]And if you say yes or you say no,
[00:02:35.700]it's actually gonna call the person,
[00:02:37.891]like send text to the person.
[00:02:39.958](upbeat music)
[00:02:41.430]Eventually when we started like researching when we found
[00:02:45.648]that using a radar signal was better,
[00:02:47.821]so from there we just got an group together
[00:02:50.923]and started researching (indistinct).
[00:02:55.355](upbeat music)
[00:02:56.996]We're currently limited on time,
[00:02:58.899]on how much we can spend on a project
[00:03:01.397]for hopefully a future capsule team
[00:03:04.428]or future research team (indistinct).
[00:03:08.040]We really wanted to do it as something
[00:03:09.949]as a proof concept so
[00:03:10.890]that it can be done and then let people have more time
[00:03:14.220]and more investment into it.
[00:03:15.840]More than just a capsule project.
[00:03:17.730]For us, we're trying to get a degree,
[00:03:19.650]hopefully someone that is really, really,
[00:03:21.120]really invested in this kit will take it somewhere,
[00:03:22.980]market it and try to save lives.
[00:03:24.660]I fell about 400 times for it
[00:03:26.340]because we had to build up all the data.
[00:03:27.840]I got really good at falling really is what I did.
[00:03:31.808](upbeat music)
[00:03:32.641]This is project Quick Grade.
[00:03:33.930]We're team 124 and our main purpose is
[00:03:36.830]to have a base station or a rover
[00:03:39.180]to aid construction degrading on job sites
[00:03:41.850]to get accurate longitude/latitude elevation data
[00:03:45.780]to minimize human error when trying to find the correct spot
[00:03:48.810]to dig or filter on a construction site.
[00:03:51.960]Over here we have our base station
[00:03:54.540]and then we have our rover set up over here, the antenna,
[00:03:58.835]and then we have our app demonstration over here
[00:04:02.070]that connects to our rover and displays PDFs
[00:04:05.820]and then the measurements taken.
[00:04:08.520]Yep, so it'll tell you who's taking your measurement
[00:04:10.830]and then the height of the rod that the rover's sitting on
[00:04:14.760]and then the blueprint will come up, everyone can see
[00:04:17.850]what the benchmark elevation's supposed to be.
[00:04:20.130]You click measure and then it'll tell you longitude,
[00:04:23.220]latitude and elevation and then who took the measurement.
[00:04:26.375]There still needs to be some more development done
[00:04:31.440]after the fact.
[00:04:32.273]This is just the prototype,
[00:04:33.870]but he provided the budget and we stayed within that
[00:04:37.020]and then he gave us design requirements and we met those
[00:04:41.250]and he's gonna sign off his approval
[00:04:43.950]and he talked to the professors and everything
[00:04:46.710]and he's really excited to use it and keep improving.
[00:04:50.210]He was trying to figure out how to get the GNSS data
[00:04:55.830]because the data is transmitted between the GNSS model
[00:05:03.300]and the satellite,
[00:05:04.133]and I sort of have to find a way to intercept that message
[00:05:08.311]and then find what the data that I'm looking for,
[00:05:13.200]look at specific messages.
[00:05:14.820]Also have to make sure that the data is specific enough
[00:05:19.350]because it says it's measuring longitude,
[00:05:23.100]usually it's not accurate all the way to an inch.
[00:05:26.550]So I have to set it into a mode
[00:05:28.530]where it can measure to the inch.
[00:05:30.611](upbeat music)
[00:05:34.200]So we created the sustainable
[00:05:36.791]small house indoor air quality monitoring system.
[00:05:39.840]It's for Dr. Chen's sustainable small house projects
[00:05:43.560]that they're building here just south of campus.
[00:05:48.150]Indoor air quality is important for the occupants' health,
[00:05:50.970]so we needed assistance to monitor such thing
[00:05:53.970]as particulate matter levels, carbon monoxide levels,
[00:05:57.996]sulfur dioxide and ozone levels
[00:06:01.350]and all those things are used to create an air quality index
[00:06:05.370]similar based on the one from the EPA.
[00:06:08.715]So if you ever see on the news when they give a weather
[00:06:11.130]forecast, they have an air quality index up there.
[00:06:14.190]So that's the same type of thing that we're calculating.
[00:06:16.290]But we're doing an active monitoring system.
[00:06:18.750]So we created three modules
[00:06:21.090]that will be in the three main rooms of the house,
[00:06:23.400]the basement, the utility room, and the main living room.
[00:06:26.700]And it will collect, constantly collect data
[00:06:29.250]and report it back to the home CR automation system.
[00:06:32.460]Which if someone will eventually completely automate
[00:06:36.270]the house and will make decisions to
[00:06:38.220]when there's a certain bad air quality inside
[00:06:40.823]and it's worse than the one that's outside,
[00:06:44.700]it'll be able to turn on a ventilation unit
[00:06:46.230]that'll swap the air and so forth.
[00:06:50.180]So it's just will record data for 90 days.
[00:06:52.980]So not only does it record live data,
[00:06:56.160]it records over a period of time so that any research
[00:06:59.550]that needed to be done or they can calibrate it later
[00:07:03.270]with the stored data.
[00:07:06.673]There's similar items on the market,
[00:07:08.100]but they only kind of record particulate matter
[00:07:10.920]or just one or two contaminants to get that AQI.
[00:07:15.900]We have all of the listed contaminants on the EPA site
[00:07:19.890]as well as a few others,
[00:07:21.180]which we can make decisions based on the external data
[00:07:24.180]as Sean was saying and what we're getting from each room.
[00:07:29.190]But along with the other contaminants,
[00:07:32.370]we also have that little element of control
[00:07:34.590]for our ventilation unit.
[00:07:35.640]Most other systems from our research hasn't shown
[00:07:40.343]that level of control.
[00:07:42.330]Was certainly a unique challenge
[00:07:44.811]that we had on this project
[00:07:45.644]was some of the sensors that we have.
[00:07:46.740]The communication protocols that we used
[00:07:48.390]to interact with microcontroller,
[00:07:50.970]they kind of like did their own thing with it.
[00:07:52.200]It was within the limits
[00:07:53.940]of those communication protocol standards.
[00:07:56.880]However, I feel personally they weren't very detailed,
[00:08:00.030]very well in the data sheet.
[00:08:01.170]So that took a lot of digging to actually find out,
[00:08:03.930]so we could get good data from the sensors
[00:08:06.150]and communicate with them properly.
[00:08:08.013]When we got the first particulate matter sensor working,
[00:08:13.650]we had spent at least two weeks without it working.
[00:08:17.655]The first problem we realized is our board
[00:08:21.750]had mirrored the connections so we had the connection swap
[00:08:26.370]and we actually had it shorted so it was getting really hot.
[00:08:29.460]Luckily it didn't damage the sensor.
[00:08:32.580]And then once we figured that out we were able
[00:08:34.380]to communicate it with it.
[00:08:35.580]But any of the data it sent back was completely incorrect
[00:08:40.320]using the software drivers that I wrote
[00:08:42.120]because I couldn't figure out how
[00:08:43.410]to integrate the ones that they provided.
[00:08:46.260]We were eventually able to get with their support
[00:08:48.150]and figure out where we had gone wrong
[00:08:50.610]and when it finally was reporting back good values
[00:08:53.310]and giving us data, it was one heck of a relief
[00:08:56.610]because that was kind of our primary sensor
[00:08:59.670]'cause that particular matter is usually
[00:09:04.127]the most likely problem to have
[00:09:08.730]for air quality indoors.
[00:09:12.720]It'd be nice to be able to get it installed
[00:09:14.250]in an actual location and see how it does,
[00:09:18.750]test it for quite a while,
[00:09:20.970]make sure that everything is still functioning
[00:09:23.505]properly after a while.
[00:09:25.379]But we definitely had some issues with our board,
[00:09:27.720]our initial board design.
[00:09:28.710]So obviously we'd want to go back
[00:09:30.900]and correct all those mistakes
[00:09:32.700]and then probably we would,
[00:09:35.070]we decided that we probably would switch out
[00:09:36.840]some of the sensors for-
[00:09:38.760]It would be nice to be able to repeat the process
[00:09:41.580]to make version two.
[00:09:44.745](upbeat music)
[00:09:46.290]So our project is the Localized Home Assistant.
[00:09:48.780]It is essentially, the closest type of product would be sort
[00:09:51.600]of an Amazon Alexa or a Google Home.
[00:09:54.270]The reason why this project was sort of designed
[00:09:56.880]and why we came to it was our sponsor Matthew Boading wanted
[00:09:59.580]a similar kind of device,
[00:10:00.720]but that was a little bit more secure, had its own network.
[00:10:03.150]So our device has no connection to an ISP,
[00:10:05.640]no internet access at all.
[00:10:07.230]Host its own local area network and webpage on it.
[00:10:09.960]You can actually see here is
[00:10:11.584]the webpage that it's hosting right now.
[00:10:13.074]You can add things like flight schedules to it,
[00:10:15.737]add new smart devices that it'll be able
[00:10:18.090]to control the timers on it.
[00:10:20.857]Additionally, it has things like voice control as well as
[00:10:26.220]like we already said, the ability to control smart lights.
[00:10:28.920]The type we're using in our design right now are Sengled,
[00:10:31.710]but it should be able to use any.
[00:10:33.750]That's sort of the basis for our design.
[00:10:36.118]And-
[00:10:36.951]The idea was actually originally from our sponsor,
[00:10:39.300]Matthew Boading.
[00:10:40.140]He wanted something that was like a home assistant,
[00:10:42.510]but more secure like Gavin was saying.
[00:10:44.460]He just basically pitched us the idea for,
[00:10:47.003]he wanted something a little bit more secure
[00:10:48.150]and then we just talked with him back and forth
[00:10:49.380]and came up with the idea ourselves
[00:10:50.610]on what parts we would need,
[00:10:52.140]how we would go about designing it.
[00:10:53.940]Things like voice recognition, what would we utilize
[00:10:56.160]for us it's a open source software called Picovoice.
[00:10:59.541]In terms of hosting like a site, how would we do that?
[00:11:02.393]Well an ESP32 chip is what we decided on that
[00:11:05.207]'cause it's able to host a local area network as well
[00:11:08.400]as an observer on that local area network.
[00:11:11.910]And then that was sort of the process was
[00:11:13.890]breaking down the project into a couple of blocks
[00:11:15.870]that were important for us
[00:11:17.610]and then finding solutions for each of those
[00:11:19.873]and then learning how to mesh those all into one project.
[00:11:22.860]Yeah, there were, I mean a lot of challenges,
[00:11:25.710]both software and hardware.
[00:11:27.990]This project is definitely really software focused
[00:11:30.480]because a lot of the requirements
[00:11:33.230]and specifications that our sponsor wanted are more
[00:11:34.980]on the home automation side
[00:11:36.540]and there's a lot of software that goes on
[00:11:38.130]with automating things like Devon said with the website.
[00:11:41.460]We have timers that you can set so you can say,
[00:11:43.027]"Hey big brother, set a timer for three minutes."
[00:11:45.390]In a sense, on the very high level that input is very easy,
[00:11:48.360]but when you get down to it, there's a lot of complications
[00:11:51.000]that come with a custom PCB
[00:11:52.830]and writing code to be then run on your custom PCB.
[00:11:56.100]And so we ran into problems with that.
[00:11:59.340]We also had a power supply system that we designed
[00:12:03.810]and initially it had about a 10th of a second delay
[00:12:08.220]and that would erase a lot of our important data
[00:12:10.590]that we wanted kept on it.
[00:12:12.330]That's volatile data like website data
[00:12:14.550]and the processor, we couldn't have it reset.
[00:12:17.340]And so especially with the light schedule that we also have,
[00:12:20.580]that's another home automation setting,
[00:12:22.350]automated home lights, setting a schedule, weekly schedule,
[00:12:25.830]and so make making sure that we go
[00:12:28.800]with the correct design that can achieve
[00:12:31.020]the sponsor requirements is essential.
[00:12:33.900]So that's why we went with a UPS,
[00:12:36.180]uninterruptible power system,
[00:12:38.373]and that definitely helped make our design
[00:12:40.440]a little bit more concrete, more professional.
[00:12:43.140]Yeah, we testing pretty extensively right now.
[00:12:47.100]Like we said, we're hosting this local site currently.
[00:12:50.040]I mean that's happening live off that thing right now.
[00:12:52.140]We have been able to control lights before;
[00:12:53.610]we don't have any year right now,
[00:12:56.062]but we have been able to do that.
[00:12:58.353]Things like the timers and all the website operations all
[00:13:01.500]work fundamentally including our UPS,
[00:13:04.020]it should be able to lose power and be good.
[00:13:06.480]The only examples of things that are out there right now,
[00:13:10.140]the closest thing we found is a Home Assistant Yellow.
[00:13:12.930]It's essentially a Raspberry Pi on its own thing.
[00:13:18.270]They have their own open source software
[00:13:19.410]that allows them to do control lights and stuff.
[00:13:22.410]It's pretty sleek.
[00:13:23.243]But that design, the concern
[00:13:24.390]with that is the design is very open source.
[00:13:27.960]It's still like if anybody can see how the design works,
[00:13:31.080]anybody can mess with it.
[00:13:32.610]So because it's very open source, it's a little less
[00:13:35.387]secure than what he wants.
[00:13:37.410]So we put in just under about 700 combined hours
[00:13:41.460]between the four of us
[00:13:42.938]and that came down from designing a PCB to mostly software.
[00:13:47.280]This was a very software intensive project
[00:13:50.340]and then it also took a total of about 70 hours
[00:13:53.370]to print all the pieces of the case to make sure
[00:13:55.230]that they were all correct and fully functional.
[00:13:59.141](upbeat music)
[00:14:03.630]So this is PLANET which stands for plant-based,
[00:14:05.700]long term autonomous nurturing eco-friendly technology.
[00:14:08.696]A bit of a mouthful,
[00:14:09.529]but how it works is it uses four different sensors
[00:14:11.940]in the soil.
[00:14:13.200]The two that we use for the healthcare functions
[00:14:14.850]are soil moisture and pH.
[00:14:16.077]And based on those with will
[00:14:17.250]automatic water and fertilize it.
[00:14:19.140]So reso senses twice a day
[00:14:20.790]and to take care of the plant for you.
[00:14:22.410]And we also have different settings within our home screen
[00:14:25.440]down here, and we can select between four different plant
[00:14:28.110]sites in the home screen right here.
[00:14:30.829]And depending on what plant type you select,
[00:14:32.910]it'll take care of the plant differently
[00:14:34.695]because a lot of plants are different.
[00:14:35.528]So be able to water it less frequently
[00:14:38.284]or more frequently depending on the plant.
[00:14:39.540]Then we also have a light attachment
[00:14:41.010]because we're partnering with Dr. Bestul
[00:14:43.200]and his research on light.
[00:14:45.180]So I don't know too much of the details of it,
[00:14:48.360]but Dr. Bestul's research is about the effects
[00:14:51.570]of the physical spectrum light, specifically blue
[00:14:54.930]and red on plants
[00:14:57.360]because we all know that plants want UV light
[00:15:00.570]but his theory is more around
[00:15:05.130]that the plants need the physical spectrum light more
[00:15:08.490]than they need UV light.
[00:15:09.657]And so that's why we chose these light colors
[00:15:12.647]as I believe we explained,
[00:15:14.580]the colors shift throughout the day and it starts at 80% red
[00:15:18.000]and 20% blue in the evening
[00:15:21.120]and then 20% red, 80% blue in the morning.
[00:15:26.550]So through the morning to night it's gonna
[00:15:28.350]shift from one to the other.
[00:15:30.210]Just gradually every hour.
[00:15:31.710]And then we also have different settings depending
[00:15:34.232]on what the light level needs,
[00:15:35.638]we can have higher intensity or lower intensity.
[00:15:38.701]It makes it a little brighter, yeah.
[00:15:42.187]Another big thing that we wanted
[00:15:44.580]to focus on was sustainability.
[00:15:46.050]So we used a lot of recycled materials
[00:15:48.476]and we have a very low energy consumption.
[00:15:51.300]At the end of the day we're using five watt hours per hour.
[00:15:54.360]So it's very low compared to a light bulb
[00:15:57.330]and where most of our packaging is recycled.
[00:15:59.550]So we're making less of an impact
[00:16:01.890]on the environment while trying to grow plants.
[00:16:04.890]That was our mission.
[00:16:06.240]There are other products on the market
[00:16:08.520]that do similar things,
[00:16:10.260]like I think the Aero Garden is probably the most popular.
[00:16:12.990]It's a hydroponic system.
[00:16:14.843]We wanted to grow plants in substrate, so in soil
[00:16:19.350]and we also wanted it to be able to grow bigger plants.
[00:16:22.050]Most of the planters like this on the market can only grow
[00:16:25.590]small plants, succulents, things like that.
[00:16:27.930]But we wanted to be able to grow potentially
[00:16:29.640]vegetables, any kind of garden herbs, flowers
[00:16:32.370]and mainly house plants.
[00:16:35.070]The other thing that the other products don't do is that
[00:16:39.150]all they do is take care of the plant
[00:16:40.800]but they don't tell you the status of it.
[00:16:42.990]And so ours, you can actually look at all of the data
[00:16:48.120]that each sensor takes and that's part
[00:16:51.446]of Dr. Bestul's research.
[00:16:53.580]That is what we need to do to make sure
[00:16:55.440]that we are taking into account the effects
[00:16:58.040]of the visible spectrum light.
[00:17:00.270]And taking it to market.
[00:17:01.410]Well I think the prototype as it is,
[00:17:03.700]would have to be developed a little more
[00:17:05.670]to make it mass consumable.
[00:17:07.890]Like we would have to rethink our packaging design
[00:17:10.040]to make it easier and we would also want
[00:17:12.656]to make it more accessible to customers.
[00:17:15.780]So driving that price down a little bit.
[00:17:17.820]So probably, I mean we did this in eight months,
[00:17:21.030]we could probably make it a much better system
[00:17:23.520]in eight more months then we could probably put out there.
[00:17:26.700]Because this is just the first interaction.
[00:17:27.803]This is the first prototype.
[00:17:29.460]That's kind of the whole point of the project.
[00:17:31.020]We're not getting a chance to go come back and redesign.
[00:17:33.355]It's just the first edition.
[00:17:34.409]Maybe another group will take it on.
[00:17:37.380]I mean I more than happy if I see a improvements
[00:17:40.049](indistinct) yeah.
[00:17:41.621]This was initially my idea, a part of it.
[00:17:45.540]The basis was my idea.
[00:17:46.620]I've had it for years
[00:17:47.640]and I think your capstone project is a perfect way
[00:17:50.190]to find an excuse to build it.
[00:17:53.006]And then Dakota's idea, 'cause Luke
[00:17:55.800]and I knew we wanted to be a team.
[00:17:57.630]Dakota was added to our team and his idea was very similar
[00:18:01.050]but it was more of a focus on larger plants
[00:18:03.690]and plants that can grow fruits
[00:18:06.379]or vegetables, things like that.
[00:18:09.990]So we kind of in the middle,
[00:18:12.398]initially it was gonna be like another smaller system
[00:18:13.890]and then we came in, expanded it
[00:18:16.110]and added all the sustainable things
[00:18:18.180]and wanted to make it a more robust system.
[00:18:21.054](upbeat music)

The screen size you are trying to search captions on is too small!

You can always jump over to MediaHub and check it out there.

Comments

0 Comments

Capstone II Spring 2024

Description

Searchable Transcript

Comments icon comment

Related Channels

Comments