Andrew's Summer Log Book (Summer 2017)
reverse chron order -- newest at top
Tuesday, June 27
This morning there is some work being done in the lab (on the roof and things) and we are unable to be in the lab to work. I think I will take the morning to try to CAD the crawler mechanism parts. I'm not sure how I will add the flexibility of the carbon fiber into my model, but I am willing to learn. Besides that, I hope to continue construction of the crawler mechanism. I should definitely try to take some more pictures and add these to this wiki this evening...stay tuned. I hope to complete building the frame today. I think that the other parts of the construction, such as attaching the spines will be a little more difficult. So, for today, we have:
- Create a mock-up of the crawler mechanism in CAD.
- Continue construction of the SuperSCAMP crawler mechanism.
- Take pictures of the progress.
Monday, June 26
Today is an exciting day! Drew is back and we will begin the construction of SuperSCAMP! Today my tasks are as follows:
- Cut out pre-dimensioned carbon fiber pieces for the crawler mechanism on SuperSCAMP.
- Build out as much of SuperSCAMP as is possible without receiving all the parts that were ordered on Friday.
- Talk to Drew and Alessandro about the timeline for the project, prepping for ICRA and getting ready for the Living Machines demo.
I talked with Drew and Alessandro for about an hour after lunch and it was a very productive meeting. We are meeting with Mark tomorrow to discuss our progress so we wanted to be ready. We also just wanted to have a good idea of what our next steps were. Drew got up and drew a fairly detailed schematic on the board (see figure below). From the schematic, we found that the construction of SuperSCAMP would boil down to 3 essential parts:
- The chassis, which was bought and is basically already complete, pending some designed and 3D printed motor mounts that Drew will make soon.
- The crawler mechanism, which will be attached to the chassis on the rotor side.
- The Crazyflie and its supporting materials such as dampers, batteries, and z-ranger (distance sensor).
Alessandro has done a remarkable job with the software implementation to control the Crazyflie and will continue to work on improving it, as well as beginning to design the takeoff approach. Drew will develop a CAD model of the chassis, the motor mounts, and the Crazyflie attachments. This will give us the ability to estimate the inertia in CAD and use it as an input into Alessandro's control model. I will be working on the construction of the crawler mechanism. I will be basing the design on previous crawler designs and simply scaling it for the SuperSCAMP's dimensions.
Today, I sized a new motor for the extend/retract motor that is used to move the spines up and down enabling SuperSCAMP to climb. Since SuperSCAMP is about 6 times the mass of SCAMP, we felt that it would be necessary to change motors to get the desired torque. After some torque analysis and motor selection (luckily one of the servos in the lab worked perfectly), I continued building the mechanism frame with carbon fiber. I plan to continue work in this frame tomorrow. One of the points that came up in our meeting was that we wanted this frame to be modular and be able to be detached from the drone. I also hope to develop a CAD model of the frame to complement Drew's work. The construction will be held up by the in/out motor, which we are expecting later this week.
We are also presenting what we are working in this Friday at the lab meeting, so I hope to have some good materials by that point.
Friday, June 23
Another late entry (Mon. 6/26):
Friday was a good day. We had a lab meeting and it was good to hear Mark talk about our goals for the summer because it gave me a concrete vision for the SuperSCAMP project. Here, I will write out some of my notes from the meeting. One of our goals is to get a paper written for an ICRA conference. The deadline to submit a paper for review for the conference is September 15th and the deadline to submit for RA letters is Sept 10th. Thus, we want to have a drafts for RA letters ready by September 2nd and drafts for ICRA ready by September 7th to give Mark enough time to review them and make edits.
For the paper, we want to focus on SCAMP's failure recovery on outdoor surfaces. The idea is that we want SCAMP to be able to detect when it doesn't have a successful perch on a surface and be able to recover from that. So, SCAMP might fly to a wall and attempt to perch. If it is unable to perch immediately, it will try to provide more power to it's rotors in an attempt to engage the spines. After a specified amount of unsuccessful attempts, SCAMP will then disengage from the wall/takeoff and hover by the wall, waiting for the next command.
We want to get this working by the end of July in order to collect some data for the paper. We need to get some data that answers some of the following questions that reviewers may have: How does SCAMP detect that it failed? What is the state machine? How many times is it good for SCAMP to try to perch before it takes off?
An idea for the data representation in the paper was that we could create a plot with lines for each of the different surfaces that graphs the probability of SCAMP's success as a function of the number of trials it tried to do so. I thought it would also be a good idea to explain our algorithm in detail.
Lastly, I purchased the parts that we needed to build the crawling mechanism for SCAMP: Loctite, accelerator, in/out climbing servo motor (HK-5320), and the extend/retract motor (HS-5035HD). They should be here hopefully by the end of the week and next week at the latest.
Thursday, June 22
Late entry (Mon. 6/26): I didn't do too much on Thursday. I was in staff training again but spent a little time in the evening emailing Will and Drew to continue to get oriented to the lab. I got a few housekeeping things done like getting after hours access to MERL and finding out which parts we needed to purchase, after talks with Will and Drew. I also continued to think about the takeoff mechanism and I'm convinced that it will be difficult to develop a reliable takeoff approach that isn't software implemented. That being said, it will be difficult to develop the software approach but I am up to the task.
Wednesday, June 21
Today, tomorrow, and possibly Friday, I will not be in the lab because I am in a training session for my role as an RA in one of the campus dorms over the summer. I am working remotely here while we are in the sessions and I will continue working remotely tonight. I will also try to see if I can get into the lab to experiment with cutting the carbon fiber later. Thus, goals for today are:
- Transfer money into my account to purchase Crazyflie (personal goal).
- Continue reading about Crazyflie and the documentation.
- Continue reading about SCAMP.
- Brainstorm ideas for the takeoff mechanism after reading part of SCAMP paper about current takeoff mechanism.
I ended up choosing not to purchase the Crazyflie right now. I think I will just send an email to Alessandro to see if he can give me a Crazyflie to work with this summer. I think that it will be very important to get some hands-on experience with the Crazyflie while I'm learning about it in building SuperSCAMP. I read some more documentation for the Crazyflie and I am becoming more confident with what I'll be able to do with it.
I continued reading the original SCAMP papers, and gained a lot of insight about the takeoff mechanism. The original takeoff mechanism for SCAMP included a takeoff arm with a single spine. This takeoff arm can be seen in the figure below derived from the SCAMP paper ("A Multimodal Robot for Perching and Climbing on Vertical Outdoor Surfaces"). The mechanism operates by releasing this takeoff arm when the climbing motors are in a certain orientation and once the climbing spines are released, SCAMP begins to fall until the spine on the takeoff arm engages an asperity. This causes the robot to pitch backward from the wall.
After thinking about this more, I believe that a software/hardware approach in altering the motor circuit to allow for driving the rotors in reverse for short periods of time would be the best way to develop a takeoff strategy. I will begin to look into ways that we can rewire the the Crazyflie to run it in reverse. In the worst case scenario, we could easily add a mechanism that pushes out an arm that creates a moment in order to get the robot to pitch backwards.
Tuesday, June 20
Welcome to my blog! I think the way I will organize this blog is by writing at two different points in the day. I will write when I come in the morning and again in the evening. In the morning, I will detail my goals for the current day and I will discuss any reflections from the previous day that may have formed while I slept or completed more work that evening. In the evening I will discuss the work that was completed that day and things that still need to be completed.
Today, is my first official day in the lab. It has been a crazy last couple of days with graduation happening and family members being in town. But I'm excited to get started! I will be working on the SuperSCAMP project this summer. This is basically a revamp of the SCAMP project to create a more robust MAV that will fly, perch and climb as the original SCAMP did. Our goal is to make SuperSCAMP less "fuzzy" and more reliable, although that may mean adding a little more weight than the regular SCAMP had. From our morning meeting, I think I will be focusing on the perching and crawling mechanism design as well as just helping out when I'm needed. Below are my goals for the day (in no particular order):
- Begin reading and tinkering with the Crazyflie documentation/software.
- Create a four-bar linkage design that we can propose to Mark for the design of SuperSCAMP's legs/feet.
- Solder broken components that Alessandro gave me.
- Continue reading papers about SCAMP.
Edits/updating after a later meeting with Mark, Alessandro, and Will (3:00 pm):
In order to make sure that we have everything we need for a demo at Living Machines (July 25th) and the big DARPA demo in August, we've decided to work in parallel to create both the same perching and climbing mechanism for SuperSCAMP as SCAMP, as well as begin to think about ways to develop a more robust mechanism for it. For the next few weeks, we will be focusing our efforts on re-building the old SCAMP climbing mechanism. We will begin construction once Drew gets back on Monday, but for now, I will be doing some preliminary work, so we can hit the ground running when he returns. I will, then, be focusing on the following tasks for the next few days, in addition to working on the tasks above as time permits:
- Dimensioning the new carbon fiber parts that will go onto the SuperSCAMP
- Looking into the specs of the motor we are currently using in comparison to other servo motors in the lab to see if it can be improved.
- Developing a more robust mechanical takeoff mechanism. Alessandro might also try to figure out a way to do this with a software strategy.
In the back of my mind, I will also be thinking about the "inchworm" wall climber. This might be something we work on later in the summer.
I completed a preliminary drawing with dimensions of what SuperSCAMP's crawling mechanism will be. I also began looking into the servo motors and decided I want us to try the HS-45HB, that is in the lab instead of the already used HS-5035HD. The reason for this change is that the HS-45HB has a higher rated maximum torque range (14-17 oz/in) when compared to the HS-5035HD (11-N/A oz/in) (see datasheets). From what I've learned in the short time talking with Mark and Will, more torque seems to be needed from the motor, as well as more robust turning, which we hope a higher voltage supply might address. I also began looking into the Crazyflie documentation. I kind of want to buy one for myself and I probably will.
I have not began to think about the takeoff mechanism yet, nor have I soldered the parts for Alessandro (one of the parts needs Molex connectors that I was unable to find so I will revisit this tomorrow). I've begun working on everything else.
I will not be in the lab for most of the day tomorrow but will continue work remotely dimensioning the crawler, reading more Crazyflie things, and reading about the takeoff mechanism in the paper again. I also plan to come into the lab to cut some carbon fiber later in the day. Hopefully, I can get in.
Monday, June 19