Yesterday was presentation day. We went to Kualoa Ranch, our research site host. Kualoa Ranch is a humongous property where the 2.7 square mile Ka’a’awa Valley is just a part of the property. On the other side of the valley is their headquarters where we would meet in the morning before heading into the valley where did our research. We met there yesterday at one of the facilities to give individual presentations on our research, 15 of us gave 10 minute presentations to an assorted audience of students, ranch staff and ownership, and community members. I was worried that we would all go over our time and bore the life out of the audience, but everyone did a really good job condensing our research into 10 minute chunks and I think the event went really well.
After we gave our presentations, we gathered for a group photo and headed to dinner. That night, we blew off some steam at the barn.
Today, we’re cleaning, packing and heading to the airport. We have some time off in the afternoon before heading to the airport, but we can’t go too far, so we’ll probably just hang around here or go to the local beach. So, that’s it, back to my normal life, actually I can’t wait, I miss it.
Yet another work day at the barn today. Tomorrow, we present our results, so today was all about the finishing touches. Namely, PowerPoint. I spent the day in PowerPoint building a short presentation on what the problem is, why it’s important to address, what the current technology provides and doesn’t provide, and how I tried to address that. My project results in a 3D model, so in my opinion screen shots won’t work. A picture of a 3D model doesn’t convey that it’s really 3D. You’ve got to see it moving around to realize it’s 3D. So, I spent the morning trying to find a good video screen capture tool. It took me 3 different tries before I found something that could adequately capture smooth quality video of me moving Google Earth and other 3D models around. I needed those videos in my presentation in order to convey what I had done. I finally got the videos using FRAPS which is intended to capture video games, but it worked for what I was doing because 3D modeling is like video games in that it requires extra horsepower. So, after all my work in PhotoScan, here is the final result:
Day 24 was Wednesday, another work at the barn day. I spent the day wrestling with PhotoScan. I’ve got around 200 photos of the cliffs and they’re loaded into PhotoScan. PhotoScan has a number of procedures that need to be run against your photos in order for it to complete its magic act of turning a bunch of flat photos into a 3D model. The first procedure is called “Align Photos”. Depending on the quality level you run it at, it can take anywhere from 30 minutes to 2 hours, for the amount of photos I have. I’ve learned that running this procedure on “High” is worth it because it creates the 3D point cloud that future steps rely on. So, I spent hours waiting for it to do its job there. Then you have to run “Build Geometry” which creates a 3D mesh out of the point cloud and creates the 3D surface for which your photos will be textured onto. This step has the potential to take the longest. I have an i7 processor with a good graphics card and an SSD drive, basically, I have one of the baddest laptop computers currently available, I just got it before this trip. Even with all that awesome hardware, it still takes my computer over 3 hours to complete this next step, even on the “Low” quality setting, that’s some hard core math. After that, you have to run the third procedure called “Build Texture” which applies the photos to the 3D models creating an amazing three dimensional model of your subject area. So, after hours of processing and trial and error, I labored to produce this 3D model. For now, I’ll show you what you get after the first step “Align Photos” which is what’s called a sparse point cloud (if you want to see the final product, read day 25):
Today was a work at the barn day. I stayed back to finish up various things, one was to catch up on my blog, another was to work on my project. I’ve been using PhotoScan with all of the images of the cliffs to create a 3D model of the cliffs. Once I get a good model of the cliffs, I can use PhotoScan to export a DEM (Digital Elevation Model). The trick is the first part, getting a good model. The software needs a lot of photos and it helps if those photos are geotagged so it knows where in space the cameras are. I had a pretty good model of the cliffs going, so I decided to export a DEM to see what it’s like. It matched up pretty well with the current terrain model that is publicly available. However, mine started at -50 meters in altitude and went to 233 meters. Unfortunately, I know that’s really wrong, because I know that our base camp is at 29 meters and where we were standing while flying was at least 10 meters above that. So that’s a problem that needs solving. My project started as a comparison of different methods of creating a 3D model, orthorphoto and DEM of vertical terrain. I’m now going to still create those same products, however, I’m going to try to merge that data with standard ground data to make a unified model of varying terrain. For now though, it’s off to bed.
Back to the field. I spent the day flying the Phantom quadcopter getting up close and personal with a particular section of cliff. I flew up and down the cliff side at varying angles with a color camera and a near infrared camera. Paul also flew with me, flying the thermal infrared camera. Mostly, the results of my flights were for Jeanette’s project, though I will be able to use the color photos in my project. In order to get close to the cliff, we had to do some hiking, so we tried to stay there without making trips back to base camp, but we ended up making a few trips anyway for different cameras or mounts. Anyway, it’s a difficult thing, flying the cliffs, with the color camera, we shoot a shot every 10 seconds, so I have to fly very slowly. The problem is, I’m flying vertically, which means I have to ascend and descend very slowly, which is very hard because you can’t really tell if you’re ascending or descending once you’re a few hundred feet away. You also have different winds to deal with, at low altitudes next to the cliffs, the wind tends to be very turbulent, at the higher altitudes, it smooths out, but tends to be an updraft. Anyway, it’s a challenge and I like that. We finished the day a bit early as I was on kitchen duty that night, so I went back to the barn and had to cut up Costco chickens, something I often do at home, only this time I had to cut six. Then I had to clean, finally, I had time to write a blog post (2 days late), then I crashed out.
My "favorite" cliff
This is the cliff, it has a cave in the upper right area that is rumored to have a burial in it.
So, we had the day off on Sunday. I was invited to go along with a group renting a car so that we could be free to do whatever we wanted. So, we caught the bus to the rental place which happened to be at Turtle Bay resort on the north shore. It was beautiful, if I ever come back to Oahu and I don’t rent a house, I could see my family staying there. Anyway, we hung out there for a bit while we waited to get our rental, but once we got our minivan, we started heading south.
Turtle Bay Resort
Turtle Bay Resort
We drove for a while before we started seeing new things as we had to back track a bit, but eventually we started checking out some sights. We saw a beach from the road that looked really nice so we stopped to check it out. It was really cool, it was almost completely local, and really not crowded. As soon as we stepped onto the sand, we noticed a turtle swimming along the shore, only 2 feet away, it was awesome. After seeing the turtle, I decided to check out what else I could see, so I snorkeled around for a while. It wasn’t the most abundant place I’ve snorkeled, but it was cool because the coral was beautiful. In most snorkel spots the coral gets pretty beat up from all the people touching and standing on it, but this place was in pretty good shape.
First fish of the day
We only stayed at that beach for about an hour and headed out, shortly after that we came upon a vista point at Waimanalo Beach Park and we stopped to take some pictures, it was an incredible view. After taking in the view and snapping picture after picture, we headed to see the Halona Blowhole, but the surf wasn’t very high, so that was kind of anticlimactic.
Waimanalo Beach Park
After the blowhole, we stopped at the Kona Brewing Company for lunch. Some of the others really wanted to go there and get souvenirs, but to me it was really nothing more than an overpriced BJ’s that wasn’t as good. After eating, we were underwhelmed by the restaurant and headed out to see more of the island. I had expressed a desire to go snorkeling at a good spot and on Oahu, that spot is Hanauma Bay. It’s an old crater that is a wildlife preserve. It’s become so popular that they charge a nominal 2 bucks to get in, and they limit how many people get in, so you have to get there early. So, I pretty much knew I wasn’t going there and sure enough, when we drove by it was sold out already, after all it was a Sunday and we were there around lunch time. No big deal, I had already been looking up alternatives on my phone while we drove and I had decided that Sharks Cove in north shore was the place to go. So, I convinced the crew to go back to north shore. From where we were near Honolulu, we took a scenic routh through the middle of the island and got to see the Dole plantation and some coffee groves and a few other interesting spots.
Within an hour we were back in the north shore area. The guys, Howard, Peter and myself decided to snorkel Sharks Cove while the girls went to do some shopping in the north shore area. For an hour and half we got to explore the cove, it was pretty amazing. We saw lots of different fish and crustaceans. While exploring, we found a huge rock on the ocean floor with a hole in it. We investigated and found that you could swim under the rock and come out the hole. As soon as I got down that deep my head felt like it was going to implode, so I didn’t go for it, but Howard and Peter borrowed my fins and made the swim, it’s in the video.
After snorkeling, we headed back to the barn to change and we headed back to Honolulu and saw the night life there. We were all pretty tired at that point so we only stuck around for a couple hours, we got our souvenir shopping done and headed back to the barn to crash for the night. So, we saw a lot in one day, and pretty much went around the island twice. I have to say, it was a really good day, we got to see and do a lot of things and we managed to keep the stress level down.
Day 20 was Saturday, yes I’m way behind. I don’t know how it got away from me, but it did. Anyway, on Saturday we had the morning to work on our projects and we had an informal discussion with the professors to go over where we were at with our projects. I spent the morning working on my data for the cliffs and running different processes with PhotoScan. I discussed my project with the professors and for the most part we agreed that I was on a good track. That evening we had a birthday celebration for Cole and we made a late night trip to the beach. It was really crazy, going to the beach so late, and the water was still warm enough for swimming.
My first late blog post :(. Yesterday, I got home later than any of the other students, ate, talked to my wife and crashed out. I woke up at 2:30 in the morning and realized that I hadn’t blogged and couldn’t get up and do it, so here I am, a day late blogging, that means I’ll have to do it again tonight.
Anyway, yesterday was crazy. I was a bit overwhelmed and got a little disgruntled at one point, but after the fact, I’m left really stoked. I showed up to the field and Iam was there with his X8 waiting for me, and immediately asked me for a flight plan, so I got busy designing a flight plan for the right side of the valley since last time I worked with him we imaged the left side. The mission I designed worked pretty well until the last turn where we turned towards the cliff and Iam decided that was no good so he aborted the flight. I definitely understood, but was a little disappointed that all of my flight plans to date had to be aborted near the end. So, after that flight was done, it was on me to get the images from the flight and properly record the flight info and catalog the images. Then I had to create another flight plan for the next flight. This time I designed the flight to turn away from the cliffs, fixed some of the altitudes of our waypoints and I checked it with Iam and he approved. The flight was 10 miles long with the UAV traveling 2 miles into the canyon and back out a number of times. By the time we got to take off, almost everyone had left the field, it was 4 PM and there weren’t many options for someone to launch the plane. I had the plane in hand already as I was checking the cameras, so I got to throw it. My heart was racing, I had already watched someone way more experienced than me throw their plane into the ground, then there’s the chance of cutting your hand or arm on the prop as it flies over your shoulder, so yeah, I was nervous and excited. I threw the plane and it headed towards the ground and skimmed the grass by about a foot, then it took off and went on it’s way to following my flight plan. I was so relieved, it was a really close call. The flight plan worked beautifully, all my turns were away from the cliffs, my altitudes were right, it all worked. On our last leg of the flight, we were about a mile away from landing and we had horses and ATVs in front of us and rain approaching from the ocean, I started to worry that we were going to have to abort again, but just in time the horses left, the ATVs drove off and we landed just as it started raining. I ran to the field to get the plane as I watched sheets of rain approaching and got to cover just in time. We finally flew a whole flight plan, all 20 waypoints, all 10 miles, it was AWESOME!
Ready for launch
I spent the whole day in the field, managing UAV flights and data, it was a lot and a bit overwhelming and super awesome. I feel so fortunate to have met Iam, he’s a Hawaiian Airlines captain flying the Airbus A330 and he’s got a few homemade UAVs. He was a really cool guy that was very collaborative and helpful. He put a lot of trust in me, even though I wasn’t experienced, he put his plane and equipment in my hands (literally) and we worked together to complete the missions. The experience working with him was truly something I won’t forget and has been one of the best things about this trip. I’m looking forward to keeping in touch with him.
After we said our goodbyes, I took an opportunity to grab a panorama of the valley:
Today I went to the study area again. I continued to get photos of the cliff walls surrounding the valley in an attempt to demonstrate how to map the cliffs. By taking photos with the quadcopters, I can put the images into some software called PhotoScan and it eventually builds a 3D model out of your photos. So, since the cliffs vary widely in angles, it’s important for me to get photos of the cliffs from a lot of different angles. To that end, I flew the quadcopter imaging the cliff walls even more today. For our first few flights, we had the camera mounted the same as yesterday which was basically on the back of the quad, so it was really difficult to fly because I had to fly backwards, which means when I pushed forward on the stick, the quad would go back, and left the quad would go right. Flying backward next to the cliffs just added that much more difficulty, but I liked the challenge. However, in the afternoon today, another group took our quad and camera to work on another part of the valley, so we had to rig up another quad. This time we were able to figure out a way to mount the camera on the front of the quad. I took the opportunity to take a quick video of the base camp area.
After that, we headed out to get more imagery of the cliffs. Everything seemed to be going really well, I wasn’t flying backwards anymore and I thought we got a lot of good pictures. However, after returning to base camp, I checked the images and a lot of them were blurry. I thought about it and we had the camera mounted very tightly to the quad, which seems like a good idea, but it causes a lot of vibration and I think that contributed to our blurry pictures. Next time we’ll try to mount the camera with some more foam and more loosely.
Just when I thought it was time to leave the study area and head back to the barn, I got the opportunity to fly a different quadcopter. This one is called the InstantEye. It has three cameras built in and a screen on the remote that lets you view any of the cameras. So, it’s like playing a video game, you watch the screen on the remote instead of watching the copter. It’s designed for the military and the company is trying to make them available to the public or at least academic institutions. You can tell it’s military, it’s very utilitarian, as it has altitude hold and a number of other useful features. We used it to get really close to one of the caves on the cliff wall, this cave is rumored to have a burial in it. We got right up close and got some good pictures, but we still couldn’t tell if there was a burial or not. Either way, it was fun flying it that close to the cliffs.
Today I started working on my individual project, which is to image, map and create a 3D model of the cliffs surrounding the valley. Although I don’t plan on getting all of the cliffs done because my individual project is about the method of mapping and modelling a vertical landscape, so I can concentrate on a section of the cliffs. I started out working on a flight plan to image the cliffs, but found that it was really hard placing waypoints that go up instead of forward or backward. The software used to deal with the arduino pilot is mission planner, and I got a lot of experience using it the other day when I was making flight plans for the planes. However, Mission Planner is designed with a Google map or any other top down map, so it’s really hard to place waypoints when the are right on top of each other. So, I gave up trying to create a viable flight plan and decided to try to fly the cliffs manually with one of the quad copters. We were able to mount our Pentax Optio to the phantom quad copter at an angle sot that it was pointing out instead of down and I felt like I got pretty good imagery.
Before I flew the cliffs for systematic imagery we took video of a flight going to the top of the cliffs, here it is, pretty spectacular :