Most of us can appreciate that flying a multicopter above packed stands is wrong. But to ban an Elev-8 from flying anywhere on a 420-acre campus seems silly and stifling. Laws and liability limit so much of what we can do nowadays. Thankfully, the middle-of-nowhere still exists. That has been the only solution I've found for the insanity we've created.
This turned out longer than Moby Dick, so skip it unless you are really into the topic.
UAS Educational Program - Stream of consciousness suggestions: apologies for repetition and unedited rhetoric and composition.
I havent tried the ELEV-8 but have fooled with 2 cheapos and can well picture the class setting based on scores of classes Ive taught with WAM & BoeBot.
Above all else: success will not be in the hardware. It will be in the teaching materials that support the program. The time requirements in computer science to develop new courses every few years already makes the field difficult and inefficient. Instructors need to step into an existing learning ecosystem. Teaching with the BoeBot can be done after an instructor does one run-through of the text; all the preparation was done by Plax. The Whats a Microcontroller SIC is the gold standard.
The SIC for this has to have a lot of instructor input and a few trials in classes before roll-out. Ive discussed many aspects of this with Andy in context of Prop C.
Break SIC into chapters that are roughly equal for completion in time. I need to plan the semester with specific objectives for each of the 15 meetings.
Courses are built around lectures & labs. Consider the materials for the lectures as much as for the labs. Most teaching materials now come with PPTs. I hate them because at end of lecture the students waking up leave slobber on the lab bench, but I think existence of PPT deck increases most instructors confidence in adopting the program.
The ELEV-8 will be the "first try" for a dozen people each semester. First tries at flying are pretty hard on the hardware. I would put lots of research into the result of dropping the craft about a hundred times from a meter high - my guess of what a couple semesters would be like. Perhaps the angled legs could be hinged at bottom so landing pressure pushes top of outer leg out the arm and can encounter some type of shock absorption.
60-second field replacement of the five parts that break most commonly. Nothing worse than first student busting it up in first five minutes of first outdoor lab.
30 second replacement of battery. Batteries with fast charge option from car. Battery connector easy to use (I have one with a tiny connector that requires exact placement of thumbnail).
Can E8 send back telemetry on flight data? Med students look at hundreds of fake EKGs to learn to ID patterns. E8 students could also have sample data with specific situations to ID. Include telemetry analysis in each chapter. Post sets of telemetry quizzes on the web site. Encourage students to post their data for comment.
Ability to use E8 in a standard PC-based flight simulator. Id require students to pass skills test there before real flight. I think simulators allow easy development of a plug-in for a new type of craft so this may be quite easy.
SKUs that are staged: basic kit at minimum price then level 2, level 3. That allows purchase in budgets over several years. Better for Plax to sell 3 minimal platforms in first year rather than that same money going into 1 tricked-out platform. If 3 basic units bought in first year then the level 2, 3 packages are sure orders in subsequent school budget cycles. Also may be able to require each student to buy a basic unit and share the labs level 2 & 3 materials.
Milestone tests indoors. Very first class should have some experiments with quantifiable results that can go into a lab report. Something like a lab to test airflow from one motor or something like a motors power consumption analysis. Perhaps some kind of lifting power test with one arm hinged to a test stand. I think anything that can break down the control into a 1- or 2- axis test first will be important. PID is probably handled in Prop code but having a section that requires writing simplest PID code in a non-flying experiment would be a useful milestone (even if not implemented in final product).
Labs that can be done indoors during spring terms Feb-Mar until outdoor labs are possible.
Must have easy & cheap way for video add-on as that is #1 cool use of UAV. Also use any other cheap & easy sensors. (I dont know the ELEV-8 platform) so students have data with which to work. Ambient thermo. Could also add thermo against battery. Is there a way to sense updrafts, perhaps change in altitude when motors steady? Sets of side-looking light sensors that can ID the horizon and thus give a pitch value.
[begin Web ideas]
Create specific web site and use every tool in the book to make the site the primary place for interactions of all things ELEV-8. It will take years of natural growth but here is how some sites jump start. Intervals below are quarterly with anticipation to increase to monthly as site builds.
Repository of data. See my servo calibration data that comes from any student in any school into an on-line database with hundreds of points. That feeds the real-time graphs. http://www.boebotteacher.com/Calibration-ServoData.aspx
Photo of the quarter contest, akin to HS best physics photo of the year.
Hack of the Quarter
Lab report of the quarter, judged by a panel of faculty. This would be an outstanding example of presenting the hypothesis, materials & methods, etc. Instructors could then assign study of that page for homework.
Section for teachers on tips, quizzes, extra projects.
Quarterly articles on writing good code from an expert: Take a sample program and show 5 ways to make it better.
Encourage teachers to have students write a piece for the site. Create proper web reference and it can be added to a students resume as a low-level publication.
Link of the quarter: link to E8 mentioned on another site.
When I say quarterly maybe adjust intervals to match up with typical semesters
[ end Web site ideas]
[begin modes]
Some useful modes in available software: (again I dont know ELEV-8, maybe these are in there already)
Auto-take over switch. Cant really have a kill switch on copter as falling is not safest option. But switch that turns control from operator to copter to immediately stabilize in place. That gives user time to collect wits.
Unstable motion take-over: Prop takes control when E8 position is changing at a rate that is probably not intended (exceeds some setting) sensitivity adjustable
Is it practical to have a self-landing option? Prop takes over landing when within 1 meter of ground
Try-it-out mode. When the head of the department comes by you want to have a mode that allows the bigwig to try it out with some success
In real planes & driver-ed cars they have a way for instructor to be in co-control while student learns, Instructor has ability to cut out student input. Is that possible?
(end modes)
Find and buy every other instructional option out there. Study them and make the Plax option better in every aspect of the teaching materials. That is what your potential buyers are going to be doing.
Educators courses in STAMP and BoeBot were very useful in building confidence for instructors to adopt the BoeBot system. Perhaps hold similar. To increase attendance you can hold before/after other conferences. Consider the current requirement to have 20 participants reduced, even if price is higher. Perhaps course could be coordinated with Plax staff existing travel (Texas Teachers, DC Science Expo). It can be frustrating to us on the old coast to see cool & free seminars in Rockland but not the equivalent back here.
In some fields (oceanography) schools will work together in a consortia. I dont know if there is some application of the model to UAS. In East there are many locations where a half dozen schools are within an hour and the faculty know each other. Perhaps a UAS consortia that offers a inter-semester course (last two weeks of May or Winter break) and rotates the location. As grow, the course may include a Plax staffer for 2 days as Engineer in Residence. Perhaps Faculty from other schools considering the ELEV8 could easily attend prior to their decision of platform. I could see a teaching assistant for the course coming from grad school of a regional powerhouse (Carnegie Mellon in our area). Another activity of consortium would be quarterly meetings with speakers and short student presentations; Amer Chemical Society is good at this. From youtube it looks like many engineering courses require students to do oral present of their project. I could see professors encourage the best to then present at the consortia meetings. Also good place for student papers/posters at BSc level.
Sometimes STEM students feel they have less opportunities than athletes and artists. My sons rugby team always went to Bermuda over spring breaks, my daughters orchestra toured Italy. The science club did neither. What about some kind of an event in FL or AZ during Spring Break? Include welcome to science groups / schools not yet using the E8. Also a good opportunity to show off the platform to a professor considering adoption.
I would make two sets of materials (SIC texts) that are geared to HS and college. The HS geared toward what can be done wo/ calc and then the university level with calc. This could be done with less writing by having the SIC for HS. Then each chapter has an extension section that brings in several more complex ideas.
Create a suggestion for syllabus timing to use materials in an inter-term course (4 weeks of half-days in Jan).
Create super-safe set of hardware for an alternate configuration to usual hardware. Objective is to allow for first trials or use in a gym or before experimental flights. Allow performance to drop in this configuration. Foam props? Extensions to arms beyond motors with ping-pong ball termini? Take a look at all safety options on other platforms and add to this configuration. Wild idea: a geodesic frame of material a little stronger than soda straw around the whole E8 (just frame, not panels). Air would flow through with little effect on flight. If it hit anything it would bounce off, including walls, ground and people. Maybe easier to make joints if not geodesic but rather grid like earth lat/long lines.
In SIC first 1-2 chapters are on board alone with some basic experiments in uCon.
Create a relationship with instructors that encourages them to write extra units and share. I think the key is to set it up as a formal location like the OBEX. Provide a template and set of guidelines. I know we can share lessons on the forum but by doing some organizing it becomes a more supportive resource. For example, Ive been working on a little extra chapter for my Automation in Art students on using a shift register that will be used in an optional seminar. I can put it on the Ed Forum but doubt it will be used. If there was an obvious, indexed place then I think instructors would take a look at it when planning the term. There seems to be more college students (incl MSc) with specialty in STEM. For them, posting a lesson would provide an additional minor publication and entry for potential employers to see on their job app. (For this whole idea of Lesson Exchange LesEx Id like to see it set up for WAM, BoeBot and Prop Learn C.)
As an education platform, consider connecting with engineering grad students. They will be the next gen of professors. Perhaps a Train-the-Trainer seminar for them held in each region and then they are authorized to do educators courses. Perhaps support the page charges when an E8-based paper is accepted. It might be easier and cheaper to develop a relationship with a group of grad students at a top-notch engineering school than to get traction with a professor. The expected payoff is in several years when the PhD candidates get jobs.
This input has been very valuable so far and I will carefully comb it for an outline before we get too far, bringing the outline back to the community for input.
I've also thought about the model we might use to develop this curriculum. We have several educators who've posted here (and a few lurkers who I've communicated with via e-mail). I'd like to collaborate with a few educators on the curriculum. We'd figure out a financial arrangement that's interesting to the educators, possibly consisting of long-term hardware at our cost so they could use it in their UAV program.
Bruce, I'd need to know if you have interest. Jason, you'd be on the hook no matter what. And Jason, what if we used the Open's 9 I/Os, Vdd, Vss to connect to a board on top that includes GPS and accelerometer? Still too much for a single Propeller? You'll be happy to know that we have Hoverfly's support to fully open this effort, too.
Milestone tests indoors. Very first class should have some experiments with quantifiable results that can go into a lab report. Something like a lab to test airflow from one motor or something like a motors power consumption analysis. Perhaps some kind of lifting power test with one arm hinged to a test stand. I think anything that can break down the control into a 1- or 2- axis test first will be important. PID is probably handled in Prop code but having a section that requires writing simplest PID code in a non-flying experiment would be a useful milestone (even if not implemented in final product).
A few years ago I made a motor test stand, and that was a bunch of fun to play around with. It's a relatively simple design, except for the torque measurement. That part could be made by Parallax, and the rest could be provided as CAD files so that students can cut it themselves with a shop bot. Or Parallax could provide life size prints, and students could cut it with a jigsaw from a sheet of plywood.
It surprising how much I learned from the thrust/torque test stand.
A few years ago, a buddy and I attempted to make a Propeller based quadrotor. That's fully documented at http://code.anzhelka.com. In particular, the final report and the description of the math needed for autonomous flight summarize the work. We got it to the point where it could balance 1 axis test stand, at which point we discovered that the CHR IMU drifted. That, and a deadline, ended further development.
We built the Anzhelka quad out of the orginal Elev-8 files that Ken posted, and cut by W9GFO. I've always wanted to revisit that project with a new IMU, and Propeller C++. Maybe that time is approaching again.
Another thing that was very helpful was our roll/pitch test stand. That's shown in this video here:
With that stand we were able to test all of our algorithms in a safe environment, without risk of damage to ourselves or our equipment. Again, something that Parallax could consider selling or making plans for.
For preliminary labs (pre-full assembly and pre-flight) the current/voltage sensor could be used for some lab projects: http://www.parallax.com/product/29130.
*** warning - use latest libraries from Adafruit or Plax. With original software this unit failed when measuring load of a motor because the first milliseconds of start there is large current draw and unit resets.
In 2D this is similar: https://www.youtube.com/watch?v=xBMKG0GZvUc
Easy to mount with simple one-bolt going down to apparatus universal joint.
Example of another safety guard to add around perimeter
It would be nice in testing to have a temp sensor attached to body of motors
I have been following this thread with a lot of interest. I think it is a great idea. Students would find it very exciting, and it is just what STEM education is all about. I discussed the idea with a lot of my students -the general reaction is eyes light up followed by a dreamy wow. John made a lot of solid education related suggestions. I have also discussed it with a few administrators and received positive reactions.As a teacher, my concerns are cost and safety. For a small class of 16 kids, I would probably have them work in groups of two and need eight copters. To pay for the the copters alone I would either need a generous grant from somewhere, or get the total cost for eight down to $3k or so.Several have mentioned prop guards. That is a minimum safety requirement. John's suggestion of a kill switch would be very helpful, even if it caused the copter to crash. It would be better than seeing it fly into the operator or spectators who would naturally gather to watch.It would also be helpful to establish safety standards, similar to the model rocketry safety code (National Association of Rocketry) , that are supported by technical experts, and by insurance companies who do business with school districts. As a teacher, I would find some comfort ( and liability protection) by rigorously following established safety practices. Science and voc ed teachers are all accustomed to this idea.
Incorporate use of oscilloscope in every chapter in the supplements for college-level study. For the HS version include only as an appendix with some simpler scenarios.
Of course, use PropScope for examples and screen shots.
Use the PropScope's DAC card so students become familiar with it instead of leaving at bottom of box.
Like telemetry data, have on web site PropScope screen shots with something wrong (or to be calculated) as exercises. Include upload of screen shots for student participation and discussion of situations
Does PropScope have a logic analyzer card available?
Could Plax joystick module be hacked to attach between E8 and test stand to provide x & y pitch (tilt) values? Probably not strong enough.Maybe possible in one-motor test scenarios.
I have been following this thread with a lot of interest. I think it is a great idea. Students would find it very exciting, and it is just what STEM education is all about. I discussed the idea with a lot of my students -the general reaction is eyes light up followed by a dreamy wow. John made a lot of solid education related suggestions. I have also discussed it with a few administrators and received positive reactions.As a teacher, my concerns are cost and safety. For a small class of 16 kids, I would probably have them work in groups of two and need eight copters. To pay for the the copters alone I would either need a generous grant from somewhere, or get the total cost for eight down to $3k or so.Several have mentioned prop guards. That is a minimum safety requirement. John's suggestion of a kill switch would be very helpful, even if it caused the copter to crash. It would be better than seeing it fly into the operator or spectators who would naturally gather to watch.It would also be helpful to establish safety standards, similar to the model rocketry safety code (National Association of Rocketry) , that are supported by technical experts, and by insurance companies who do business with school districts. As a teacher, I would find some comfort ( and liability protection) by rigorously following established safety practices. Science and voc ed teachers are all accustomed to this idea.
Lev, would be fantastic to have you on board.
Your input about a kill switch is an important suggestion, too. We could simply use another radio channel for this purpose. If there is no radio contact, the ESCs could just shut off under control of the firmware, but if there is radio connection and things are getting out of control we could use one of the radio's alternate switches. I think we have plenty of inputs on the Hoverfly boards.
I am a believer of safety, too. Provided certain practices are followed, quadcopters can be used without people getting hurt. It might take additional precautions, like a net or acrylic shield to protect a student. Safety glasses are a good start, of course.
What if we used the Open's 9 I/Os, Vdd, Vss to connect to a board on top that includes GPS and accelerometer? Still too much for a single Propeller? You'll be happy to know that we have Hoverfly's support to fully open this effort, too.
In the quad I have now, with self-leveling, here's my cog usage:
Spin:
1) Main flight code
2) DCM routine - Updates the orientation matrix @ 250Hz. Would need to be in PASM to add magnetometer to it because I have no more time in Spin. Altered some of the FloatMath code to include conversion to fixed point so I don't have to do them in Spin, which got me to @ 250Hz.
PASM:
3) RC receiver
4) Servo8Fast (modified Servo32 to drive 8 outputs at 250Hz)
5) FloatMath (modified to provide ASin / ACos using only a single cog instead of two)
6) SensorInput - reads gyro, accelerometer, and magnetometer. Could read other I2C sensors as well, but I'm running of of code space
7) FullDuplexSerial - Used for debugging
8) SD Data writing routines for logging - This actually uses 2 cogs, so I have to disable FullDuplexSerial if I want to use it
The last two aren't strictly necessary, but they're really useful. The RC receiver cog could conceivably read some other sensor in its downtime, but that'd mean a max rate of 50Hz. That'd likely be fine for the GPS or the barometer.
The big problem from here is that the barometer and magnetometer both require reasonably complicated math to incorporate them in. The magnetometer reading needs to be "oriented" using a matrix multiplication, and then folded into the DCM code. The conversion from barometer pressure to altitude is non-linear and complicated. The example code is written in Spin, and uses float math, and that'd need to be done in PASM to be fast enough, and then the flight code would need to incorporate all this new data.
I think it's *possible* to do it in one Prop, but I suspect it'll be quite challenging to write, you'll have no cogs left over, and probably not even serial output for debugging.
My idea for my next one is to have one Prop that does all the sensor & RC receiver reads, and handles the motor outputs. This Prop would talk to another Prop running all the flight code, orientation estimation, and complex math for the magnetometer and barometer. It would send back motor values for the sensor Prop to output. That would leave you with enough cogs to do data logging, serial output, and maybe have some left over for future expansion.
I've given some thought to writing the DCM code in C, which would make it faster, and easier to do some of the ugly math. That means figuring out how to make it play nice with Spin, and C takes a lot of code space to boot, especially when you start doing floating point math.
Sorry for the "core dump". Yes, likely possible, but really hard. I suspect there's a reason the Hoverfly Pro uses two Props.
I'm not so sure kill switches are the best idea. I think it would be better/safer if all quads had IMUs so that they could right themselves and descend slowly (or to a home location) in the event of loss of radio contact.
A kill switch being activated accidentally, or by panic could lead to a situation just as bad as a runaway quad.
Put another way, there may be a scenario where the kill switch would be useful but there will be accidental or unnecessary usage of the kill switch that may lead to damage or injury.
My DJI-Naza based quad has a failsafe mode (it can be switch-activated, but happens automatically on loss of radio signal) that does a "return home and land". It's saved my bacon (and my quad) a couple times now. The mode is very slow and controlled, so anyone nearby has plenty of time to get out of the way, and it first does a 60 foot ascent to make sure it clears obstacles. It's not a perfect solution, but it works well in most cases. I'd disable it if I was flying under a lot of tree cover, for example.
A straight up kill switch would basically mean "turn my quad into a falling object", which is probably not what most people would want.
When I brought up the topic of a kill switch I noted the same problem W9 saw with a true kill. I actually suggested an alternative (along lines of W9)
>>>
Auto-take over switch. Can’t really have a kill switch on copter as falling is not safest option. But switch that turns control from operator to copter to immediately stabilize in place. That gives user time to collect wits.
>>>
But no hard feelings if you didn't catch that in my post. You would have to be almost OCD about training materials to go through my long list of ideas.
Jason's input above was really well-received. We will need to make our own firmware for this project, in C, and his efforts are the primary starting point. Jason, thank you for your contributions. I'm hoping we'll have access to your contributions when we decide to proceed.
We are looking to assemble a team of contributors and educators at this stage - contributors, reviewers, and beta testers. If you fit into this category, please let it be known by e-mail to my attention or by posting here on this thread.
Send me an email and I'll give you what I have so far. My latest implementation of the DCM code maintains 28-bit precision in fixed point (probably more than necessary, but it was free).
I have code that does self-leveling (it's what I flew at the Prop Expo last summer) but I haven't done much with it for a while - I had a crash which damaged the hardware and haven't gotten a replacement together yet. Most of the PASM driver-level stuff is the same as what I'm using in the Open / Sport code, but there have probably been optimizations and some code cleanup done since then.
This input has been very valuable so far and I will carefully comb it for an outline before we get too far, bringing the outline back to the community for input.
I've also thought about the model we might use to develop this curriculum. We have several educators who've posted here (and a few lurkers who I've communicated with via e-mail). I'd like to collaborate with a few educators on the curriculum. We'd figure out a financial arrangement that's interesting to the educators, possibly consisting of long-term hardware at our cost so they could use it in their UAV program.
Bruce, I'd need to know if you have interest. Jason, you'd be on the hook no matter what. And Jason, what if we used the Open's 9 I/Os, Vdd, Vss to connect to a board on top that includes GPS and accelerometer? Still too much for a single Propeller? You'll be happy to know that we have Hoverfly's support to fully open this effort, too.
More discussion would be great.
Ken Gracey
Yes,,I am very interested in following through with this. I expect to be moving forward, one way or another, over the next two weeks
Yes,,I am very interested in following through with this. I expect to be moving forward, one way or another, over the next two weeks
For your uses at Tarrant County Community College, can you tell me what STEM criteria you'd like to see applied to the course? Specifically, will your drone education program need to show adherence to any standards like these:
I realize these apply mostly to high school levels. I imagine you'd like a curriculum which applies the base-level knowledge in a real-world application with UAVs, right? Are there any specific standards you'd need to adhere to?
Along with safety, safety, safety (did I say "safety"?), stress the importance of observing all local and federal laws concerning UAV operation, restricted areas, cargo limits, etc.
Maybe have the laws on-hand for class discussion.
As regards to restricted areas if one of these comes over my property it had better be out of reach of my 12 ga. 36" barrel goose gun shooting high brass #4s.
But it could blossom into a whole new sports outlet what with the constant decline of migratory avian species.
If you think you are a hot shot UAV pilot bring it on out to my test range and I will give you a real world hostile environment. (I keep the salvage rights of whatever falls on my property.)
Hands off the mouse - step away from the keyboard. Out
The last three posts are not relevant to the topic. To a visiting educator it could be viewed very negatively, and this statement has nothing to do with personal freedoms but our business case. At some point our program could become quite popular and the media might not understand how to communicate our efforts as they already have trouble with communication on the subject without references to hunting drones. Please hold further unrelated discussion not on the topic.
No big deal. You make significant and funny contributions around here and I know there's no bad intentions. And trooks needn't be put on alarm over it either. Worry not.
For your uses at Tarrant County Community College, can you tell me what STEM criteria you'd like to see applied to the course? Specifically, will your drone education program need to show adherence to any standards like these:
I realize these apply mostly to high school levels. I imagine you'd like a curriculum which applies the base-level knowledge in a real-world application with UAVs, right? Are there any specific standards you'd need to adhere to?
Ken Gracey
Let me take a look at these and I will also review the grant for specific information
I think a course in UAS would be very beneficial. I built an ELEV-8 HEX-copter for my Electronics Projects Class. There needs to be more information regarding configuration and binding to the transmitter receiver. I did a lot of research and sent numerous emails to tech support. I also had to contact Hoverfly since the arming sequence was not working properly. I also had to contact Spectrum for assistance with setting gains for proper sensitivity when manually flying. There needs to be more documentation on what the limitations are, of this platform. Otherwise everything else is superb.
Let me take a look at these and I will also review the grant for specific information
Brce
According to the grant, the course we design should be available to high school students. Based on this I would be inclined to set up learning objectives based on the documents you provided.
We could include math, physical science, and engineering as appropriate.
Just storing an idea on this thread presented by another educator: redundant, backup system to auto-locate a downed ELEV-8 would be a fantastic application.
The big problem from here is that the barometer and magnetometer both require reasonably complicated math to incorporate them in. The magnetometer reading needs to be "oriented" using a matrix multiplication, and then folded into the DCM code. The conversion from barometer pressure to altitude is non-linear and complicated. The example code is written in Spin, and uses float math, and that'd need to be done in PASM to be fast enough, and then the flight code would need to incorporate all this new data.
I just got my barometer going, I think. Its BMP085, but from china, and might be flakey, but it was cheap enough to experiment. It might need to be recalibrated, if thats even possible; it reads the temperature about 4C too hot. Because of this, i think, it says the pressure is a few hPa too low (997 instead of 1014) and finally this make the altitude off by something like 18 meters. All that aside, it SEEMS repeatable, so working as relative measurements instead of absolulte measurements might be ok.
Anyway, Sal did it all in scaled integer and it seems realy fast. The code is pretty small, although it is a bunch of lines of forth code.
Anybody else try scaled integer fixed point math for this? It seems like a great fit.
Comments
UAS Educational Program - Stream of consciousness suggestions: apologies for repetition and unedited rhetoric and composition.
I havent tried the ELEV-8 but have fooled with 2 cheapos and can well picture the class setting based on scores of classes Ive taught with WAM & BoeBot.
Above all else: success will not be in the hardware. It will be in the teaching materials that support the program. The time requirements in computer science to develop new courses every few years already makes the field difficult and inefficient. Instructors need to step into an existing learning ecosystem. Teaching with the BoeBot can be done after an instructor does one run-through of the text; all the preparation was done by Plax. The Whats a Microcontroller SIC is the gold standard.
The SIC for this has to have a lot of instructor input and a few trials in classes before roll-out. Ive discussed many aspects of this with Andy in context of Prop C.
Break SIC into chapters that are roughly equal for completion in time. I need to plan the semester with specific objectives for each of the 15 meetings.
Courses are built around lectures & labs. Consider the materials for the lectures as much as for the labs. Most teaching materials now come with PPTs. I hate them because at end of lecture the students waking up leave slobber on the lab bench, but I think existence of PPT deck increases most instructors confidence in adopting the program.
The ELEV-8 will be the "first try" for a dozen people each semester. First tries at flying are pretty hard on the hardware. I would put lots of research into the result of dropping the craft about a hundred times from a meter high - my guess of what a couple semesters would be like. Perhaps the angled legs could be hinged at bottom so landing pressure pushes top of outer leg out the arm and can encounter some type of shock absorption.
60-second field replacement of the five parts that break most commonly. Nothing worse than first student busting it up in first five minutes of first outdoor lab.
30 second replacement of battery. Batteries with fast charge option from car. Battery connector easy to use (I have one with a tiny connector that requires exact placement of thumbnail).
Can E8 send back telemetry on flight data? Med students look at hundreds of fake EKGs to learn to ID patterns. E8 students could also have sample data with specific situations to ID. Include telemetry analysis in each chapter. Post sets of telemetry quizzes on the web site. Encourage students to post their data for comment.
Ability to use E8 in a standard PC-based flight simulator. Id require students to pass skills test there before real flight. I think simulators allow easy development of a plug-in for a new type of craft so this may be quite easy.
SKUs that are staged: basic kit at minimum price then level 2, level 3. That allows purchase in budgets over several years. Better for Plax to sell 3 minimal platforms in first year rather than that same money going into 1 tricked-out platform. If 3 basic units bought in first year then the level 2, 3 packages are sure orders in subsequent school budget cycles. Also may be able to require each student to buy a basic unit and share the labs level 2 & 3 materials.
Milestone tests indoors. Very first class should have some experiments with quantifiable results that can go into a lab report. Something like a lab to test airflow from one motor or something like a motors power consumption analysis. Perhaps some kind of lifting power test with one arm hinged to a test stand. I think anything that can break down the control into a 1- or 2- axis test first will be important. PID is probably handled in Prop code but having a section that requires writing simplest PID code in a non-flying experiment would be a useful milestone (even if not implemented in final product).
Labs that can be done indoors during spring terms Feb-Mar until outdoor labs are possible.
Must have easy & cheap way for video add-on as that is #1 cool use of UAV. Also use any other cheap & easy sensors. (I dont know the ELEV-8 platform) so students have data with which to work. Ambient thermo. Could also add thermo against battery. Is there a way to sense updrafts, perhaps change in altitude when motors steady? Sets of side-looking light sensors that can ID the horizon and thus give a pitch value.
[begin Web ideas]
Create specific web site and use every tool in the book to make the site the primary place for interactions of all things ELEV-8. It will take years of natural growth but here is how some sites jump start. Intervals below are quarterly with anticipation to increase to monthly as site builds.
Repository of data. See my servo calibration data that comes from any student in any school into an on-line database with hundreds of points. That feeds the real-time graphs. http://www.boebotteacher.com/Calibration-ServoData.aspx
Photo of the quarter contest, akin to HS best physics photo of the year.
Hack of the Quarter
Lab report of the quarter, judged by a panel of faculty. This would be an outstanding example of presenting the hypothesis, materials & methods, etc. Instructors could then assign study of that page for homework.
Section for teachers on tips, quizzes, extra projects.
Quarterly articles on writing good code from an expert: Take a sample program and show 5 ways to make it better.
Encourage teachers to have students write a piece for the site. Create proper web reference and it can be added to a students resume as a low-level publication.
Quarterly contest of best haiku about ELEV8
Quarterly contest of Best ELEV8 videomusic mix. See this as example from our sailing club: http://discsailing.org/disc-boats/rogue/ (bottom of page)
Link of the quarter: link to E8 mentioned on another site.
When I say quarterly maybe adjust intervals to match up with typical semesters
[ end Web site ideas]
[begin modes]
Some useful modes in available software: (again I dont know ELEV-8, maybe these are in there already)
Auto-take over switch. Cant really have a kill switch on copter as falling is not safest option. But switch that turns control from operator to copter to immediately stabilize in place. That gives user time to collect wits.
Unstable motion take-over: Prop takes control when E8 position is changing at a rate that is probably not intended (exceeds some setting) sensitivity adjustable
Is it practical to have a self-landing option? Prop takes over landing when within 1 meter of ground
Try-it-out mode. When the head of the department comes by you want to have a mode that allows the bigwig to try it out with some success
In real planes & driver-ed cars they have a way for instructor to be in co-control while student learns, Instructor has ability to cut out student input. Is that possible?
(end modes)
Find and buy every other instructional option out there. Study them and make the Plax option better in every aspect of the teaching materials. That is what your potential buyers are going to be doing.
Educators courses in STAMP and BoeBot were very useful in building confidence for instructors to adopt the BoeBot system. Perhaps hold similar. To increase attendance you can hold before/after other conferences. Consider the current requirement to have 20 participants reduced, even if price is higher. Perhaps course could be coordinated with Plax staff existing travel (Texas Teachers, DC Science Expo). It can be frustrating to us on the old coast to see cool & free seminars in Rockland but not the equivalent back here.
In some fields (oceanography) schools will work together in a consortia. I dont know if there is some application of the model to UAS. In East there are many locations where a half dozen schools are within an hour and the faculty know each other. Perhaps a UAS consortia that offers a inter-semester course (last two weeks of May or Winter break) and rotates the location. As grow, the course may include a Plax staffer for 2 days as Engineer in Residence. Perhaps Faculty from other schools considering the ELEV8 could easily attend prior to their decision of platform. I could see a teaching assistant for the course coming from grad school of a regional powerhouse (Carnegie Mellon in our area). Another activity of consortium would be quarterly meetings with speakers and short student presentations; Amer Chemical Society is good at this. From youtube it looks like many engineering courses require students to do oral present of their project. I could see professors encourage the best to then present at the consortia meetings. Also good place for student papers/posters at BSc level.
Sometimes STEM students feel they have less opportunities than athletes and artists. My sons rugby team always went to Bermuda over spring breaks, my daughters orchestra toured Italy. The science club did neither. What about some kind of an event in FL or AZ during Spring Break? Include welcome to science groups / schools not yet using the E8. Also a good opportunity to show off the platform to a professor considering adoption.
I would make two sets of materials (SIC texts) that are geared to HS and college. The HS geared toward what can be done wo/ calc and then the university level with calc. This could be done with less writing by having the SIC for HS. Then each chapter has an extension section that brings in several more complex ideas.
Create a suggestion for syllabus timing to use materials in an inter-term course (4 weeks of half-days in Jan).
Create videos of specific problems and solutions for assignment to students as homework. Examples of mine: http://www.boebotteacher.com/Animation-BoeBotTurns.aspx and http://www.boebotteacher.com/Animation-EscapeCorners.aspx
Create super-safe set of hardware for an alternate configuration to usual hardware. Objective is to allow for first trials or use in a gym or before experimental flights. Allow performance to drop in this configuration. Foam props? Extensions to arms beyond motors with ping-pong ball termini? Take a look at all safety options on other platforms and add to this configuration. Wild idea: a geodesic frame of material a little stronger than soda straw around the whole E8 (just frame, not panels). Air would flow through with little effect on flight. If it hit anything it would bounce off, including walls, ground and people. Maybe easier to make joints if not geodesic but rather grid like earth lat/long lines.
In SIC first 1-2 chapters are on board alone with some basic experiments in uCon.
Create a relationship with instructors that encourages them to write extra units and share. I think the key is to set it up as a formal location like the OBEX. Provide a template and set of guidelines. I know we can share lessons on the forum but by doing some organizing it becomes a more supportive resource. For example, Ive been working on a little extra chapter for my Automation in Art students on using a shift register that will be used in an optional seminar. I can put it on the Ed Forum but doubt it will be used. If there was an obvious, indexed place then I think instructors would take a look at it when planning the term. There seems to be more college students (incl MSc) with specialty in STEM. For them, posting a lesson would provide an additional minor publication and entry for potential employers to see on their job app. (For this whole idea of Lesson Exchange LesEx Id like to see it set up for WAM, BoeBot and Prop Learn C.)
As an education platform, consider connecting with engineering grad students. They will be the next gen of professors. Perhaps a Train-the-Trainer seminar for them held in each region and then they are authorized to do educators courses. Perhaps support the page charges when an E8-based paper is accepted. It might be easier and cheaper to develop a relationship with a group of grad students at a top-notch engineering school than to get traction with a professor. The expected payoff is in several years when the PhD candidates get jobs.
This input has been very valuable so far and I will carefully comb it for an outline before we get too far, bringing the outline back to the community for input.
I've also thought about the model we might use to develop this curriculum. We have several educators who've posted here (and a few lurkers who I've communicated with via e-mail). I'd like to collaborate with a few educators on the curriculum. We'd figure out a financial arrangement that's interesting to the educators, possibly consisting of long-term hardware at our cost so they could use it in their UAV program.
Bruce, I'd need to know if you have interest. Jason, you'd be on the hook no matter what. And Jason, what if we used the Open's 9 I/Os, Vdd, Vss to connect to a board on top that includes GPS and accelerometer? Still too much for a single Propeller? You'll be happy to know that we have Hoverfly's support to fully open this effort, too.
More discussion would be great.
Ken Gracey
A few years ago I made a motor test stand, and that was a bunch of fun to play around with. It's a relatively simple design, except for the torque measurement. That part could be made by Parallax, and the rest could be provided as CAD files so that students can cut it themselves with a shop bot. Or Parallax could provide life size prints, and students could cut it with a jigsaw from a sheet of plywood.
It surprising how much I learned from the thrust/torque test stand.
A few years ago, a buddy and I attempted to make a Propeller based quadrotor. That's fully documented at http://code.anzhelka.com. In particular, the final report and the description of the math needed for autonomous flight summarize the work. We got it to the point where it could balance 1 axis test stand, at which point we discovered that the CHR IMU drifted. That, and a deadline, ended further development.
We built the Anzhelka quad out of the orginal Elev-8 files that Ken posted, and cut by W9GFO. I've always wanted to revisit that project with a new IMU, and Propeller C++. Maybe that time is approaching again.
Another thing that was very helpful was our roll/pitch test stand. That's shown in this video here:
With that stand we were able to test all of our algorithms in a safe environment, without risk of damage to ourselves or our equipment. Again, something that Parallax could consider selling or making plans for.
*** warning - use latest libraries from Adafruit or Plax. With original software this unit failed when measuring load of a motor because the first milliseconds of start there is large current draw and unit resets.
In 2D this is similar: https://www.youtube.com/watch?v=xBMKG0GZvUc
Easy to mount with simple one-bolt going down to apparatus universal joint.
Example of another safety guard to add around perimeter
It would be nice in testing to have a temp sensor attached to body of motors
Of course, use PropScope for examples and screen shots.
Use the PropScope's DAC card so students become familiar with it instead of leaving at bottom of box.
Like telemetry data, have on web site PropScope screen shots with something wrong (or to be calculated) as exercises. Include upload of screen shots for student participation and discussion of situations
Does PropScope have a logic analyzer card available?
Could Plax joystick module be hacked to attach between E8 and test stand to provide x & y pitch (tilt) values? Probably not strong enough.Maybe possible in one-motor test scenarios.
Lev, would be fantastic to have you on board.
Your input about a kill switch is an important suggestion, too. We could simply use another radio channel for this purpose. If there is no radio contact, the ESCs could just shut off under control of the firmware, but if there is radio connection and things are getting out of control we could use one of the radio's alternate switches. I think we have plenty of inputs on the Hoverfly boards.
I am a believer of safety, too. Provided certain practices are followed, quadcopters can be used without people getting hurt. It might take additional precautions, like a net or acrylic shield to protect a student. Safety glasses are a good start, of course.
Ken Gracey
In the quad I have now, with self-leveling, here's my cog usage:
Spin:
1) Main flight code
2) DCM routine - Updates the orientation matrix @ 250Hz. Would need to be in PASM to add magnetometer to it because I have no more time in Spin. Altered some of the FloatMath code to include conversion to fixed point so I don't have to do them in Spin, which got me to @ 250Hz.
PASM:
3) RC receiver
4) Servo8Fast (modified Servo32 to drive 8 outputs at 250Hz)
5) FloatMath (modified to provide ASin / ACos using only a single cog instead of two)
6) SensorInput - reads gyro, accelerometer, and magnetometer. Could read other I2C sensors as well, but I'm running of of code space
7) FullDuplexSerial - Used for debugging
8) SD Data writing routines for logging - This actually uses 2 cogs, so I have to disable FullDuplexSerial if I want to use it
The last two aren't strictly necessary, but they're really useful. The RC receiver cog could conceivably read some other sensor in its downtime, but that'd mean a max rate of 50Hz. That'd likely be fine for the GPS or the barometer.
The big problem from here is that the barometer and magnetometer both require reasonably complicated math to incorporate them in. The magnetometer reading needs to be "oriented" using a matrix multiplication, and then folded into the DCM code. The conversion from barometer pressure to altitude is non-linear and complicated. The example code is written in Spin, and uses float math, and that'd need to be done in PASM to be fast enough, and then the flight code would need to incorporate all this new data.
I think it's *possible* to do it in one Prop, but I suspect it'll be quite challenging to write, you'll have no cogs left over, and probably not even serial output for debugging.
My idea for my next one is to have one Prop that does all the sensor & RC receiver reads, and handles the motor outputs. This Prop would talk to another Prop running all the flight code, orientation estimation, and complex math for the magnetometer and barometer. It would send back motor values for the sensor Prop to output. That would leave you with enough cogs to do data logging, serial output, and maybe have some left over for future expansion.
I've given some thought to writing the DCM code in C, which would make it faster, and easier to do some of the ugly math. That means figuring out how to make it play nice with Spin, and C takes a lot of code space to boot, especially when you start doing floating point math.
Sorry for the "core dump". Yes, likely possible, but really hard. I suspect there's a reason the Hoverfly Pro uses two Props.
Jason
A kill switch being activated accidentally, or by panic could lead to a situation just as bad as a runaway quad.
Put another way, there may be a scenario where the kill switch would be useful but there will be accidental or unnecessary usage of the kill switch that may lead to damage or injury.
A straight up kill switch would basically mean "turn my quad into a falling object", which is probably not what most people would want.
>>>
Auto-take over switch. Can’t really have a kill switch on copter as falling is not safest option. But switch that turns control from operator to copter to immediately stabilize in place. That gives user time to collect wits.
>>>
But no hard feelings if you didn't catch that in my post. You would have to be almost OCD about training materials to go through my long list of ideas.
Jason's input above was really well-received. We will need to make our own firmware for this project, in C, and his efforts are the primary starting point. Jason, thank you for your contributions. I'm hoping we'll have access to your contributions when we decide to proceed.
We are looking to assemble a team of contributors and educators at this stage - contributors, reviewers, and beta testers. If you fit into this category, please let it be known by e-mail to my attention or by posting here on this thread.
Thanks,
Ken Gracey
I have code that does self-leveling (it's what I flew at the Prop Expo last summer) but I haven't done much with it for a while - I had a crash which damaged the hardware and haven't gotten a replacement together yet. Most of the PASM driver-level stuff is the same as what I'm using in the Open / Sport code, but there have probably been optimizations and some code cleanup done since then.
Yes,,I am very interested in following through with this. I expect to be moving forward, one way or another, over the next two weeks
For your uses at Tarrant County Community College, can you tell me what STEM criteria you'd like to see applied to the course? Specifically, will your drone education program need to show adherence to any standards like these:
http://www.edtx.org/college-ready-standards-and-practices/t-stem/ or
http://standards.nsta.org/Standards/AccessStandardsByTopic.aspx
I realize these apply mostly to high school levels. I imagine you'd like a curriculum which applies the base-level knowledge in a real-world application with UAVs, right? Are there any specific standards you'd need to adhere to?
Ken Gracey
As regards to restricted areas if one of these comes over my property it had better be out of reach of my 12 ga. 36" barrel goose gun shooting high brass #4s.
But it could blossom into a whole new sports outlet what with the constant decline of migratory avian species.
If you think you are a hot shot UAV pilot bring it on out to my test range and I will give you a real world hostile environment. (I keep the salvage rights of whatever falls on my property.)
Hands off the mouse - step away from the keyboard. Out
Hey, trooks, that was an April Fool's thing anyway. They don't sell depleted U to just anybody, you know.
Thanks but no thanks.
Not quite as good as what I am using already. I have more whallop and longer reach.
Depleted Uranium is only good in shaped charges because it turns to plasma under extreme pressure.
I am using what are known as 4" magnums with a full 2 oz. of shot.
Using a slug instead of shot these are elephant stoppers.
Thanks,
Ken Gracey
Okay, sorry, Ken. Maybe trooks can edit his, too.
No big deal. You make significant and funny contributions around here and I know there's no bad intentions. And trooks needn't be put on alarm over it either. Worry not.
Ken Gracey
Let me take a look at these and I will also review the grant for specific information
Brce
According to the grant, the course we design should be available to high school students. Based on this I would be inclined to set up learning objectives based on the documents you provided.
We could include math, physical science, and engineering as appropriate.
http://www.modelaircraft.org/membership/membershipoptions.aspx
Ken Gracey
I just got my barometer going, I think. Its BMP085, but from china, and might be flakey, but it was cheap enough to experiment. It might need to be recalibrated, if thats even possible; it reads the temperature about 4C too hot. Because of this, i think, it says the pressure is a few hPa too low (997 instead of 1014) and finally this make the altitude off by something like 18 meters. All that aside, it SEEMS repeatable, so working as relative measurements instead of absolulte measurements might be ok.
Anyway, Sal did it all in scaled integer and it seems realy fast. The code is pretty small, although it is a bunch of lines of forth code.
Anybody else try scaled integer fixed point math for this? It seems like a great fit.