Asked — Edited

Live Hack Ideas/Suggestions

Hey everyone. It’s the time of year I start hosting live hacks. I’d like to know your thoughts on subjects to cover. I have a list below and will add to it with your feedback. What are your thoughts?

Hacks subjects - SSC-32

  • servo pcb control hbridge
  • rock pi x
  • Servo Camera server with unity
  • navigation/slam discussion (what sensor options are available, what do people want out of navigation, etc)
  • exosphere telepresence option
  • making a robot skill
  • making a movement panel
  • tensor flow & yolo object detection
  • IPS with Glyph
  • intel realsense tracking (T265)
  • neat-o botvac lidar hack
  • databot review


Related Hardware Rock Pi X

ARC Pro

Upgrade to ARC Pro

ARC Pro is your passport to a world of endless possibilities in robot programming, waiting for you to explore.

#1  

Hi DJ,

All good ideas to me. Im for the telepresence:) and that yolo object detection looks interesting.

PRO
Belgium
#2  

robot skills like making a walking gait ?

#3   — Edited

I like Nomad's idea!  I do have one question about EZB brand HD servos. Everytime I buy one from Robotshop Canada they send me the servo with no plastic gear to actually use it on my robot,I try hot Glue to attach,no good, Krazy glue--works for about a week then lets it go. I saw you one time using a dremmel to shape the motor shaft into a slotted gear. Before I try something extreme like that is there any place I could buy some plastic gears to fit my HD servos? DJ or Jer any suggestions?

PRO
Canada
#4  

I would like to see functional robot that does something useful.  Example a robot that changes the Keurig container in my coffee machine, puts a cup underneath and pushes the button so I have fresh coffee when my alarm goes off.  Or a robot that can play a simple card game like SNAP so you put down a card and it puts down a card and if they match says snap before you. Oh and of course the infamous get me a beer bot.

If I have to pick from list tensor flow and yolo (but I guess you could use that for the card playing bot as well or the I said I wanted a Corona not a Heineken beer bot)

PRO
Synthiam
#5  

Nink - i won't be building a robot, that's what i make the software for:D so you can build the robot. I'll demonstrate different programming techniques and solutions to problems you're attempting to solve.

@Nomad & Roborad, there's dozens of Auto Position tutorials demonstrating how to use it. I don't think that needs to be covered any further. Between The Robot Program episodes and the Auto Position manual, there's dozens of videos.

Any ideas of what programming challenges or technology integrations you'd like to see? (outside of building robots)

#6  

Right on Nink! I played the PC version of Fallout 4 and there is a robot that dispenses beer and just follows you around so you can shoot enemy and have a beer to celebrate the successful attacks.

#7   — Edited

The newer Roomba and some Chinese knock off roombas do map the house and remember where to go clean, I have seen some new plugins that may be able to do this in synthiam Arc? I think it needs to use wheel encoders to work.

PRO
Synthiam
#8  

The new roomba isn't very hackable and too expensive for people to use in robot projects

The old Neat-o vacuums have built-in 360 degree lidar that is quite hackable. That would be a fun hack

As for other navigation things, the IPS is something that everyone has dismissed considerable. I'll add that to the list to demo how it works.

And another navigation system i haven't spent much time with but plan on making a skill for is the intel realsense tracking camera

#9  

Ya true that 700.00 for a new Roomba forget about it, The Chinese versions though are only about 250.00 and can map rooms in memory,better than shark? Not sure but I may buy one for Christmas to see.

PRO
USA
#10  

I’d love to see anything dealing with navigation. Like the Neato lidar and or the realsense camera. I’d really like robots to map rooms and avoid obstacles and wall.

Portugal
#12  

RealSense Depth Camera would be my choice. As i said almost 2 years ago, i think it is the way to go in robot navigation. Its low price, free sdk and an almost one in all solution makes it THE choice for personal diy robots. For example the included IMU makes the use of encoders unnecessary for SLAM.

#13   — Edited

Botvac lidar hack would interest me too!

#14  

I would vote for lidar or something with navigation. Try to keep it low cost and let us know far enough in advance so we can divert some Christmas money if needed.

#15  

Just checking if anything tonight? Just bought and downloaded the new starwars Squadrons as it was today 40 percent discount with graphics that are just mind blowing.Although still want to see a good hack if anything going on.

#16   — Edited

I am on Star Wars Squadrons, lets meet up...:D

For a life hack I would also be up for anything indoor navigation related, I always wanted to build a little tracked robot to explore my appartement!!:)

#17  

Amazing space fighter simulation but I am still having a tough time getting used to all the different ships control behavior,all so different, The Tie Fighters are too sensitive to my joystick,always spiral into asteroids,LoL and R2 will never shut up with my erratic x wing flying!

#18  

Wow just heard the true British actor at 80 just died he played Darth Vader in first 3 starwars movies,he was in the suit but voice was done by American James Earl Jones.

#19   — Edited

Hey DJ which old Neato Vacs have the Lidar? Looked at the D7-8 on amazon very expensive and no good customer reviews,you are likely a 1 in 3 chance of buying a defective new unit,1 guy got 2 more peplacements,both broken,gave up asked for money back.....Then again it may be really cheap to buy used if has something defect but easy to fix and has working Lidar! EDIT--- On further investigating Amazon Canada, I did see the Neato D4 which still had Lidar but still price of 560.00 CND...I can buy a Lidar cheaper to just add on my Create 2 Roomba hacks.

#20  

Hey Ezang that is a cool Link,looked into the real sense 3d Cam, and it uses on the fly intelligence to quickly identify objects in the room using an almost SCI-FI sensing program similar to how "Predator" Alien seeks target and modify vision to target enemy! Demo here

Portugal
#23  

Yep, the T265 is best for vSlam.

PRO
Canada
#24  

The T265 disclaimer looks interesting.  WARNING This product can expose you to Antimony oxide (Antimony trioxide), which is known to the State of California to cause cancer.

Portugal
#25  

So do other 100.000.00 of things....:(

PRO
Synthiam
#26  

Is that in the plastic? Or a sensor? Weird that it only causes cancer in the state of California:D haha love that wording

PRO
USA
#27  

....yeah california has a very long list, exhaustive at times.....

Quote:

California's cancer-warning policy comes from a 1986 state law called Proposition 65, enacted to protect California's drinking-water supply from toxic and potentially cancer-causing chemicals. It also mandates the state keep a master list of all chemicals known to be toxins and requires manufacturers and businesses to warn people about these chemicals if they're present in products or buildings, even in extremely small doses.

PRO
USA
#28   — Edited

@DJ: Regarding the camera T265, it seems the product is on auto pilot mode, is not discontinued or retired and is available, but, Intel moved the focus to other products.

I have been burned a few times by intel. I have two realsense cameras deprecated on a short period time, broken promises e.g. skeleton tracking sdk. Looking to the other side: embedded boards: I have 2 edison, 1 damaged Galileu and 1 x Arduino Intel curie somewhere in the basement, everything is gone , remember Intel Joules (EZB 5).

It's my guess, and off course I'm biased not very happy with some Intel decisions, but, you can check the open tickets and won't be fixed replies, plus the lack of response. There are a little bit of everything: stability issues, SDK crashes, missing features, and broken promises.

It's a shame because it's really a good idea.

PRO
Synthiam
#29  

yah - i'm in the same opinion. The main trouble is there isn't a good plug'n'play navigation system out there. The IPS is great for indoor but no one is manufacturing it. And doing any inside out navigation like the realsense is.. well, that's all there is.

I'm hoping because there are enough real sense around that it'll at least stick around for a bit. It's not very well supported - i had a call with the cto of realsense division and even he didn't seem super optimistic about it, which was a little disappointing. I do recall our effort with Intel during the EZB V5... we had great expectations but they used us for the dog and pony show on stage and then discontinued the product.

on that note, it isn't a great SDK to work with. There's a lot of marketing buzz words about navigation, localization and SLAM... but it's not really there. It can give you some data about movement but it doesn't know where you are. It also doesn't seem to create a map of where you are or how to get anywhere, so no path generation.

Not sure exactly how anyone else is using it. Hoping someone will shed some light

PRO
USA
#30  

@ptp do you have a good suggestion for an affordable solution for navigation ( inside and out) if not intel products?

PRO
Synthiam
#31  

Wonder if we're going to have to group together and make something --- or make something compatible with something that already exists.

@ptp, the EZ SLAM skill will take lidar data and generate a map. However, the issue we ran with that is dead reckoning would drift. It needs wheel encoders. And even with wheel encoders, we found the SLAM map eventually started twisting and getting bent out of shape because the wheels slip.

So really the SLAM is good for immediate room navigation but not good enough for a whole house or anything - in a real world environment. I know there's test cases and laboratory/university examples of slam working super well. But in the real world, we're not seeing it often because it's really used for obstacle avoidance.

I'm not sure what a real good navigation system is. Maybe we should start by discussing our navigation needs/requirements and working from there.

PRO
USA
#32   — Edited

James Bruton has a turtlebot 3 and I think he used ROS navigation, but in that particular video he says its lidar had to be used with wheel encoders....so must always be drifting (slipping) issue when using lidar and lightweight robots...at least for the robot locating itself with in the map accurately...also explains why my Neato has very thick silicone wheels!

So no Lidar and no TOF (time of flight) cameras..what options does that leave?

#33  

The roombas come heavy with good traction wheels and encoders,so you experts are saying that even if I buy the Cheaper 360 Lidar systems,it won't really be able to make simple maps that can store on a pc like Latte panda? I am sure they come with coding samples to use on Arduinos or Python ,slower for me to learn than Synthiam ARC samples. They are only used as obstacle avoidance? I sure hope DJ does get something that maps nicely.or else just keep using sonar.

Portugal
#34   — Edited

@DJ: I was thinking about the IPS idea and came across a cheap way of doing so, it is cald "aruco marker navigation". As i understand it, it can be used with single or multiple markers. The markers give x,y,z coordinates. Take a look here:  https://www.youtube.com/watch?v=JuyzlFzVbbE But i am still confident that the realsense is the best option for navigation. Let me know what u think guys.

PRO
Synthiam
#35  

Yeah the glyph marker can be used right now. The camera robot skill actually provides the rotation (in degrees) of the detected glyph. So it would be very easy for you to make that happen. You'd never need to "guess" the direction of the robot because it would be always known. Some simple javascript would make that work easily.

You'd be limited to the distance with the camera and a glyph, however. Mainly because the further you get the larger the glyph will need to be. In a small home room, it wouldn't have to be huge. But it would be interesting for testing non the less.

PRO
Synthiam
#36  

Oh - i should also add that glyph IPS is how 2 of synthiam's exosphere robots work. The robots have glyphs on top of them to know where they are and when the task has been accomplished. Then it uses that glyph to locate the robot into a random starting area for the next user.

User-inserted image

PRO
Synthiam
#39  

Are you asking for aruco marker detection rather than using the existing glyphs for this usage?

PRO
Canada
#40   — Edited

There is a little startup in Calgary called TakeMeTuit that do location tracking.  I met with CEO and his team and they track location using speakers and the microphone in a mobile phone.  They can get accuracy within a few centimetres in a 3 dimensional location so much better than Bluetooth or indoor magnet field detection.  I have been meaning to talk to them about using their tech to track robots locations but if someone wants me to set up a meeting I can.

PRO
USA
#41   — Edited

I like that idea of gaining a 3rd dimension (z) using a 2d glyph. That's very similar of how motion capture works in the film industry.

PRO
Synthiam
#42  

Nink, that would be neat. It’s hardware? How’s the price tag? I see users here that’d want it:)

will, what would you do with the third dimension of tilt? And it’s only a few degrees, remember. As much as the visibility of the actual image to the camera. I’d need a use case to get an idea of what the skill would look like and what features it would provide you. The third dimension doesn’t help with navigation though.

PRO
USA
#43   — Edited

I was thinking more in conjunction with the intel camera map for location (combined) ..but its not that part that fails for the mapping....im thinking of it for another unrelated use tho

PRO
Synthiam
#44   — Edited

That would be one or the other. You see, the ips system mode (like what we are discussing) has a remote camera watching the room to see where the robot is. That is absolute positioning. It’s a godlike view of your robots location:)

the intel realsense (or slam based sensors like lidar) are not that. They use a collection of sensors mounted on the robot that tries to understand its position in an unknown and constantly changing space. This is why this type of navigation doesn’t exist as a viable or reliable solution. Otherwise, if this was solved, we’d have robots everywhere. This is the problem the entire planet of all super intelligent minds are having difficulty solving - well that and conversational Turing style AI. However, it seems this post says he’s solved all of that or made breakthroughs that would make great robot skills.

PRO
Synthiam
#45  

Added SSC-32 as a hack suggestion

Portugal
#46  

Hey DJ, when are you going to do the live hack?

PRO
Synthiam
#47  

I’ll most likely do one of those Jan 15th I’ll have time

PRO
USA
#48  

I have tried the servo pcb board for position with a DC motor and I can never get it to work or blue smoke the board. Love to see a hack for servo pot and pcb to DC motor Hbridge....like a big one 12 volts or something similar. Thanks.

Portugal
#49  

Hey fxrtst, have you tried the T265 camera yet. Mine arrives Tuesday.

PRO
USA
#50  

@proteusy Not yet, worst time of year for a project...family first...hoping too after the holidays.

PRO
Synthiam
#51  

Will, did you forget to connect an hbridge to the servo pcb?

PRO
USA
#52  

Yeah, I had an Hbridge connected. It was the TRex motor controller. This was quite awhile ago when I set it up. I tried with three pcbs and pots from a couple of ez robot servos.

PRO
Synthiam
#53  

The hbridge will require logic input on the direction pins. So the servo therefore can only be powered by logic voltage (3 or 5v what ever the hbridge input logic is)

PRO
USA
#54  

...yeah it was 5V logic from H bridge. So pretty sure no blue smoke from that, unless the EZ robot servo pcb logic is less than 5 volts?! But maybe I messed up the wiring. I'll have to review it again soon. That is why I was asking to add it to hack night. I was piggy backing off Daves sabertooth thread...just wanted to post here to keep everything where it should be:).

PRO
Synthiam
#55   — Edited

I appreciate that:) (organization)

PRO
Canada
#56  

I am voting for the Neato Lidar Hack with Navigator

#57  

I 2nd Nink's vote for the Neato Lidar.

Alan

PRO
USA
#58  

Quote:

I’ll most likely do one of those Jan 15th I’ll have time
@DJ Are you going to do a Hack tomorrow?

PRO
Synthiam
#59  

No. I’m out of the province stuck in covid lock down in northern Ontario...

PRO
USA
#60  

Oh no! Ok stay safe...!!!

#61  

Ya it sux,here in southern Ontario, I am about to try to go to Grocery store, If you dont here from me again I probably got into a fight with a security guy or cop because I refuse to follow arrows on the store floor or parking too long at Tim Hortons parking lot where they shoo you away if you stay in your car to drink coffee.

PRO
Synthiam
#62   — Edited

Thanks. It’s not all bad. My family is here so visiting them during this weird time is good. Did some ice fishing last week. Had a few fish fry evenings this week. Mostly doing a lot of work on the cabin. Keeps me busy:) wish we had more snow though so I could go skidoo-ing. Worst winter for snowmobiling but I don’t mind the mild temps anyway

PRO
Synthiam
#63  

Oh my... anyone want to see a live hack on making robots come to life and be real and conscience? O_o there's no shortage of dreamers out there

#64   — Edited

DJ , I always marvel at a spider or ant on the floor, how it reacts and runs faster when my hand goes near it . It is tiny but still seems smart enough to get out of danger. I am wondering how much does that ant actually think or if it is all just sensory stimuli causing a reaction with all ant nerves wired together. This giving it the run away movement. OR is it in control by thinking Oh hell danger, get out fast! The spider must have some thinking going on to chase that ant ,captured in webbing saved for later to eat. I Mean I have used my new phone to zoom in on a spiders face, it almost looks evil and thinking with those beady eyes.xD  Wondering are we close to providing matching A.I. to a bug? Or all still just puppetry done with clever engineering?

PRO
Belgium
#65  

hi roborad

here an interesting video about future robot bugs .

PRO
USA
#66  

all I can say is  WOW

PRO
USA
#67  

@ DJ, Will you be doing anything from this list tonight? If not, do you know when you are scheduling now that you are back?

PRO
Synthiam
#68   — Edited

Not sure when I’ll jump back in. Lots of other priorities keeping ARC updated and new feature development going on:)

most of everything in the list has been done with manuals on their respective robot skill pages

PRO
Synthiam
#70  

I was thinking of doing some hacks during the week. How would you guys feel about that? Cause Friday’s don’t always line up with my schedule. Gonna do one on Tuesday and try it out

PRO
USA
#71  

For sure anytime is a good time!