ROS Engineering group messages -- check your junk/spam folder

[Michael Wimble]

I've resent the announcement for Wednesday's presentation by Sergei. If
you're in the ROS Engineering SIG e-mail list, some people are not
seeing the messages. I'm not sure why these messages often get routed to
spam/junk, but you should check to make sure you've got the announcement
for Wednesday's presentation, it should be a good one.

[Pito Salas]

Michael

Do you have a separate mailing list for the engineering group? I’d be. Interested in joining but honestly I don’t have it in me to commit to a 2hour presentation. So when and you all decide to ease that expectation, I’d be interested to join.

- PIto

> On Jul 16, 2024, at 12:44 PM, Michael Wimble <mwi...@gmail.com> wrote:
>
> I've resent the announcement for Wednesday's presentation by Sergei. If you're in the ROS Engineering SIG e-mail list, some people are not seeing the messages. I'm not sure why these messages often get routed to spam/junk, but you should check to make sure you've got the announcement for Wednesday's presentation, it should be a good one.
>

> --
> You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/hbrobotics/8cae639a-3c79-4c6d-a86a-f08fb09508ac%40gmail.com.

[Michael Wimble]

There is a separate mailing list, but I’m a hard ass about “no voyeurs”. You must contribute to take part in the group. It’s a commitment that you will take the time to give a 2 to 2.5 hour detailed presentation about your robot. Heavy on the tell, light on the show. It’s your chance to give back. And I will be very pissed if there are people on the list who don’t, over the next few months, make a presentation.

I’ve run a few organizations in my life and it always, always turns out the same. Six people do everything. Over a twenty year span in any group, it’s almost always the same six people. The organizations I manage now require participation. There are plenty of places to learn about robots, only a few which I find particularly juicy. This group is a pretty juicy group—lots of information compressed into a short timespan. With pointed questions from those who are fairly far along on engineering their robots. I imagine there are plenty of people that would like to sit in and watch. But groups that are larger than a dozen or so seldom get anything done. This is not a group that is just trying to make everyone happy.

So, you know I’d love to have you join the group. Just come back when you have the time and content to give back to the group. Hope that is soon.

> To view this discussion on the web visit https://groups.google.com/d/msgid/hbrobotics/963B67F0-E7B1-4DBC-9A91-9AF90EAA517F%40gmail.com.

[Chris Albertson]

At some point I could talk about walking theory and parameterized leg motion. But in ROS-Speak, the entire walking robot is just the “base controller” that accepts “command velocity” and “command pose” messages. This may have to change when I start to worry about feet and legs colliding with other objects and walking on uneven floors including stairs. I don’t know how that will turn out yet.

The theory I like is from biology. Animals that move, or pump blood or breath have neural structures called “cyclic pattern generators” one kind is a ring of nuerons with a pulse that cycles in a loop. Like a mechanical distributer in an old car, there are taps in the loop. THen there are inputs that tell the loop how fast, whaen ther phase should be and if the timing should ba symetric or not and so on. The loop is hard-coded by genetics but control over ther loop can be learned to some degree depending on if the loop is for walking or heartbeats. Nominally, this is what I’m working but the dependancies drill down several layers and that is where I’m stuck now, driiging a trench so that later I can lay a foundation and so on.


About opening a club to only those who are at your level, I see why. But on the other hand, I organize a bicycling group. One person just wrote and said she rides about 12 miles every week and does about 7 to 8 MPH, could I join your group?" I have to say “we do a littel more then twice your speed and distance on roads that have some good size hills." So our group rides at an intermediate level and she is a beginner. I see a problem. How to beginners raise to the level of intermediate? I don’t like the idea of saying “go and train on your own and come back after a year.” If everyone said that there would be fewer intermediates. But I have not solved this problem. You can not ride together if one is doing 8 MPH and the other 18 MPH

Major League Baseball solved this problem by inventing “farm teams”.

I retired from engineering and taught high shool science for a few years. One interresting thing they taught me while working on an masters degree in secondary education was the research done in how to break a classroom up into groups. There are two ways (1) By abilty, you place the best A-students at the same table and the C-students at their own. This is matching abilty. the other is (2) you use mixed abilty so that each table has a mix, you spread the best and worst students around and let them interact. People have experimented and kept records and by far “mixed abilty” wins, even in the case of the better students doing better. Mixed groups force more interaction and the best students will try and explain to the others, explaining is a GREAT way to learn. And if the only thing the lower abilty students do is copy the answer from the others, that is more then they would do if I put all of therm together. They all win. You might argue that like bicycling, it is best to group them by speed so that no one is held back. But not in this case.

I think 9th grade biology and bike riding might have some relevance. I’m thinking of how to organize a local (Redondo Beach, CA) robot club. I ABSOLUTLY do not want meetings where we set up chairs for an audiance and have presentations. The format will the chairs in a circle. We will have a 3-minute egg timer and a rule that says if you start talking you start the timmer and have to stop after 3-minutes and let someone else speak. I will also have projects. From years working in aerospace engineering I know that engineers are on a scale of expertise and experiance and a good manager breaks up a project and makes asignments based on abilty. The team us usually mixed abilty and different specialties. I want the club to work like that. People can joint projects and contribute as they can. I would be VERY happy to accept a graphic artest who knows nothing at at about computers. The ideas is not to emulate a university lecture class but rater a university research lab where you might have Phd students and undergrads working but of course on different projects.

[Michael Wimble]

Thanks, Chris. I appreciate all the contributions you make to the group, sharing in your knowledge. Choose a month when you’d like to give a presentation and we’ll work it out in private e-mails to get it set up.

As for the meeting format, I’ve grappled with these issues for decades, too.  My feeling is that there are literally tens of thousands of resources for people to learn robotics. Myself, I’ve designed super computers, wrote compilers, wrote AI systems, was an authority on debugging of complex systems, worked on search engines, yada, yada and learning robots is the hardest thing I’ve done. Mostly because robotics isn’t a narrow engineering field—it involves such a massive field of knowledge and skills. At 74, my ability to learn and focus is declining and I have to fight harder and harder all the time to make progress. 

As a result, I’ve decided that this is a group for the fast riders. Not every group needs to cater to every one. It’s a good goal, and often it works well. But I don’t have time to slow down the bike ride for a beginner, I need to build a robot while I can that will help take care of me as I age. I cannot do it by myself. And even with the over-thousand-members of the Home Brew Robotics Club, there are hardly two people working on the same robot at all. The best we have done is to share what knowledge we have so that others can progress faster.

I’m all for other efforts to get people on board. You haven’t been at the Tuesday evening SIGs, but every week myself and a few others are there to give back, to answer questions about ROS and robotics engineering for others at any level. But there are still only six people or so that come more than once to take advantage of all the information sharing we do there. 

I am spending an inordinate amount of my mortality trying to learn how to build a mobile platform that can roam my house and not smack into things. There are plenty of people in the club trying to make robots move about, but almost no one is trying to do it reliably, robustly and so on. It’s hard. It’s very, very hard. And it’s not because I’m lazy or stupid. 

As I often whine, just because your software tells a motor controller to turn a wheel doesn’t mean the motor controller even sent a signal, and if the signal was sent, it doesn’t mean the motor actually turned, and if the motor turned it doesn’t mean that the wheel also turned, and if the wheel turned it doesn’t mean it moved at the velocity and for the duration that was commanded, and if the wheel move at the velocity and time commanded it doesn’t mean that the robot itself moved, and if the robot actually moved, I can pretty much guarantee it didn’t move where you expected and when you expected. Nothing about robots is easy.

So, back on point, if any one at all in the club wants a more inclusive gathering, then just step up and do it. We have, as with pretty much every club I’ve been in for my whole life, about six people in the club who actually step up to make the club work. I’m involved in the monthly meetings, three SIGs, and plenty of private conversations, all while trying to build my own robot all on my own—all of it. Please, please step up, anyone who wants to make a difference.

For those who can’t make the commitment to share knowledge, don’t get upset with my group, just take advantage of the thousands of other opportunities to learn robotics. Better, start your own SIG, publish your own BLOG, join the existing SIGs, but do something.

But, again, thanks Chris for your thoughts, you sharing.

[Chris Albertson]

I don’t disagree.   I’m just saying that in general it is a hard problem and the problem is more universal then just robots.

You are right abut “if so,mething needs to be done, just step up and do it.”   I’m looking for a venue.  I might be forced to use Zoom initially.

I agree about participation,   A very efftive method, but the kids hate it, is a stack of index cards, one for each student, I would shuffle then them pull of a card at random and ask “John, Do you have any idea why...”, then I’d pull another card and ask again.   No one was allowed to hide in the back of the class and keep quiet.   With adults, I think just going around in a circle and asking everyone to say what problems they are currely stuck on would get things going.   A lot more people have things that hold them up than finished projects to show off.      I’m stil thinking about how to do this.  And again a venue that’s free.

On Jul 17, 2024, at 9:07 AM, Michael Wimble <mwi...@gmail.com> wrote:

Thanks, Chris. I appreciate all the contributions you make to the group, sharing in your knowledge. Choose a month when you’d like to give a presentation and we’ll work it out in private e-mails to get it set up.

As for the meeting format, I’ve grappled with these issues for decades, too.  My feeling is that there are literally tens of thousands of resources for people to learn robotics. Myself, I’ve designed super computers, wrote compilers, wrote AI systems, was an authority on debugging of complex systems, worked on search engines, yada, yada and learning robots is the hardest thing I’ve done. Mostly because robotics isn’t a narrow engineering field—it involves such a massive field of knowledge and skills. At 74, my ability to learn and focus is declining and I have to fight harder and harder all the time to make progress. 

As a result, I’ve decided that this is a group for the fast riders. Not every group needs to cater to every one. It’s a good goal, and often it works well. But I don’t have time to slow down the bike ride for a beginner, I need to build a robot while I can that will help take care of me as I age. I cannot do it by myself. And even with the over-thousand-members of the Home Brew Robotics Club, there are hardly two people working on the same robot at all. The best we have done is to share what knowledge we have so that others can progress faster.

I’m all for other efforts to get people on board. You haven’t been at the Tuesday evening SIGs, but every week myself and a few others are there to give back, to answer questions about ROS and robotics engineering for others at any level. But there are still only six people or so that come more than once to take advantage of all the information sharing we do there. 

I am spending an inordinate amount of my mortality trying to learn how to build a mobile platform that can roam my house and not smack into things. There are plenty of people in the club trying to make robots move about, but almost no one is trying to do it reliably, robustly and so on. It’s hard. It’s very, very hard. And it’s not because I’m lazy or stupid. 

As I often whine, just because your software tells a motor controller to turn a wheel doesn’t mean the motor controller even sent a signal, and if the signal was sent, it doesn’t mean the motor actually turned, and if the motor turned it doesn’t mean that the wheel also turned, and if the wheel turned it doesn’t mean it moved at the velocity and for the duration that was commanded, and if the wheel move at the velocity and time commanded it doesn’t mean that the robot itself moved, and if the robot actually moved, I can pretty much guarantee it didn’t move where you expected and when you expected. Nothing about robots is easy.

I think we have to make a decision.  Are we going for a good enough expedient solution or is the goal to invent new science and solve the problem.  I like to see of biology has not already solved this problem.  What do coyotes do when thery chace a rabbit.  Coyotes are good at not running into trees and rocks.  I think they use a fundamentally different navigation algorithm then do our ROS powered robots.  Robots look at the plotted path and their current location then compute the error and attempt to correct the error to zero.   Coyotes never do that.  They predict their future state from the current state and then change the current state so as to improve theor future state and so on working at several diffdrent timescales.  There is nothing at all like a PID algorithm inside the coyote.

LLMs are predictive like this.   They are trained to predict the next word, given a sequence of words and they run that in a loop.   Coyotes, I think are like LLMs that prediuct the next few milliseconds given the current state and this just runs continuously.   LLMs us language because the traing data was available but I thin “Large Action Models” are possible.  But this involves oinventing new science and if you do that the robot is never finished.

So, back on point, if any one at all in the club wants a more inclusive gathering, then just step up and do it. We have, as with pretty much every club I’ve been in for my whole life, about six people in the club who actually step up to make the club work. I’m involved in the monthly meetings, three SIGs, and plenty of private conversations, all while trying to build my own robot all on my own—all of it. Please, please step up, anyone who wants to make a difference.

I’m still thinking about how.  But I really like the university research lab type organization.  Maybe someone builds a small robot independantly but the goal is to build a test platform for some larger purpose.   You give the “build a mechanical robot” to the new person or undergrad while the more experienced person works on the control software.  Something like that.  But “project based”.    Maybe to quote Marx "From each according to his ability, to each according to his needs”.  Ok a communist lab where collectivly something gets done.

For those who can’t make the commitment to share knowledge, don’t get upset with my group, just take advantage of the thousands of other opportunities to learn robotics. Better, start your own SIG, publish your own BLOG, join the existing SIGs, but do something.

But, again, thanks Chris for your thoughts, you sharing.

On Jul 17, 2024, at 8:27 AM, Chris Albertson <alberts...@gmail.com> wrote:

At some point I could talk about walking theory and parameterized leg motion.   But in ROS-Speak, the entire walking robot is just the “base controller” that accepts “command velocity” and “command pose” messages.   This may have to change when I start to worry about feet and legs colliding with other objects and walking on uneven floors including stairs.   I don’t know how that will turn out yet.

The theory I like is from biology.  Animals that move, or pump blood or breath have neural structures called “cyclic pattern generators”  one kind is a ring of nuerons with a pulse that cycles in a loop.   Like a mechanical distributer in an old car, there are taps in the loop.  THen there are inputs that tell the loop how fast, whaen ther phase should be and if the timing should ba symetric or not and so on.  The loop is hard-coded by genetics but control over ther loop can be learned to some degree depending on if the loop is for walking or heartbeats.      Nominally, this is what I’m working but the dependancies drill down several layers and that is where I’m stuck now, driiging a trench so that later I can lay a foundation and so on.


About opening a club to only those who are at your level, I see why.  But on the other hand, I organize a bicycling group.  One person just wrote and said  she rides about 12 miles every week and does about 7 to 8 MPH, could I join your group?"   I have to say “we do a littel more then twice your speed and distance on roads that have some good size hills."    So our group rides at an intermediate level and she is a beginner.    I see a problem.  How to beginners raise to the level of intermediate?  I don’t like the idea of saying “go and train on your own and come back after a year.”  If everyone said that there would be fewer intermediates.     But I have not solved this problem.   You can not ride together if one is doing 8 MPH and the other 18 MPH

Major League Baseball solved this problem by inventing “farm teams”.

I retired from engineering and taught high shool science for a few years.   One interresting thing they taught me while working on an masters degree in secondary education was the research done in how to break a classroom up into groups.  There are two ways (1) By abilty, you place the best A-students at the same table and the C-students at their own. This is matching abilty.  the other is (2) you use mixed abilty so that each table has a mix, you spread the best and worst students around and let them interact.    People have experimented and kept records and by far “mixed abilty” wins, even in the case of the better students doing better.  Mixed groups force more interaction and the best students will try and explain to the others, explaining is a GREAT way to learn. And if the only thing the lower abilty students do is copy the answer from the others, that is more then they would do if I put all of therm together.  They all win.   You might argue that like bicycling, it is best to group them by speed so that no one is held back.  But not in this case.

I think 9th grade biology and bike riding might have some relevance.   I’m thinking of how to organize a local (Redondo Beach, CA) robot club.  I ABSOLUTLY do not want meetings where we set up chairs for an audiance and have presentations.  The format will the chairs in a circle.  We will have a 3-minute egg timer and a rule that says if you start talking you start the timmer and have to stop after 3-minutes and let someone else speak.   I will also have projects.   From years working in aerospace engineering I know that engineers are on a scale of expertise and experiance and a good manager breaks up a project and makes asignments based on abilty.  The team us usually mixed abilty and different specialties.    I want the club to work like that.   People can joint projects and contribute as they can.   I would be VERY happy to accept a graphic artest who knows nothing at at about computers.   The ideas is not to emulate a university lecture class but rater a university research lab where you might have Phd students and undergrads working but of course on different projects.

-- 
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hbrobotics/0EAB85E3-ECB5-4F8C-B158-2928622DA095%40gmail.com.

[Eva Carrender]

Hi Chris and others,

I've been reading your messages. The round table discussion group sounds like an interesting idea. You all have been discussing some really challenging issues. 

So here's how I can step up. You said you need a venue. I can offer Hacker Dojo in Mt. View as a meet-up space and would be willing to host your events. 

If you are interested DM me and we can discuss the possibilities. 

All the best,
Eva Carrender
Evacar...@yahoo.com 

Technology Information Instructor 

Making Learning Fun

Sent from my Android, please excuse brevity and typos.

On Wed, Jul 17, 2024 at 10:29 AM, Chris Albertson

<alberts...@gmail.com> wrote:

https://groups.google.com/d/msgid/hbrobotics/188454A1-C86F-4598-929D-DA8B995D6FC8%40gmail.com

.

[Chris Albertson]

>
> Hi Chris and others,
>
> I've been reading your messages. The round table discussion group sounds like an interesting idea. You all have been discussing some really challenging issues.
> So here's how I can step up. You said you need a venue. I can offer Hacker Dojo in Mt. View as a meet-up space and would be willing to host your events.
> If you are interested DM me and we can discuss the possibilities.

Thanks, except that I am in Redondo Beach, (just South of LAX, near Los Angeles). I already have a name: “Redondo Robotics”. By coincidence “Redondo” already implies “round"