Henry IX: A fully a...
 
Notifications
Clear all

Henry IX: A fully autonomous robot platform

147 Posts
14 Users
63 Reactions
18.9 K Views
robotBuilder
(@robotbuilder)
Member
Joined: 5 years ago
Posts: 2187
 

@byron 

Cheeky!  I bounce out of bed at 5am in the morning, cook breakfast for the wife and go for a nice long walk.  So far so good. Biological age vs chronological age.

More important than the number eyes is how well the program can process the visual data. We merely collect visual data with our eyes,  the actual "seeing" is a brain process,  and the same is true for machine vision.

 


   
ReplyQuote
(@jonnyr)
Member
Joined: 4 years ago
Posts: 35
 

@robotbuilder @codecage  @byron @bigbadjohn

Chasing away rabbits would be a good application in Australia. We have loads of them here. Also chasing away foxes and cane toads. Or a drone that can zap flies and mosquitos, or a robot that shoots a water pistol at pigeons on the balcony.

I think in terms of cost and technology, it feels like more advanced stuff, that was once only accessible to corporate R&D, is seeping down from industry to us hobbyists.

Another good application is autonomous mine detection bots for places like Angola and Cambodia. You could probably put together a kit with tank tacks for around 150 USD to do this. And have high school kids and university students make them.

In terms of vision, lasers combined with cameras is definitely going to allow an AI robot to have abilities that exceed natural stupidity.

Also, I think there is a lot of work on processing algorithms for machine vision, but feature extraction is still a bit under rated. 


   
ReplyQuote
byron
(@byron)
No Title
Joined: 5 years ago
Posts: 1163
 
Posted by: @robotbuilder

More important than the number eyes is how well the program can process the visual data.

Posted by: @jonnyr

Also, I think there is a lot of work on processing algorithms for machine vision, but feature extraction is still a bit under rated. 

There's been quit a lot of interesting activity in the visual processing area with AI and whatever in the last few years.  I think that having some co-processors dedicated to the visual requirements (say something like a Teensy 4) that just reports momentous events back to the main robot processor could be something to explore, especially if mounted on a big bot capable of carrying large batteries.

Posted by: @jonnyr

or a robot that shoots a water pistol at pigeons on the balcony.

I do like the idea of a well aimed squirt of water.   Not far from us a neighbour put in an automaton that move its fishing rod at intervals meant to scare the birds away, particularly heron, from pinching his ornamental carp.  After initial success the birds soon got to ignore its movements and perch on its head.  I was wondering if the rabbits would get to ignore the bot, but a squirt of water should get their attention.  Plus I could get it to squirt some visiting kiddies 😀 . That vision algorithm better work well or the bot may go and squirt me too. 😳 


   
Jonnyr reacted
ReplyQuote
(@jonnyr)
Member
Joined: 4 years ago
Posts: 35
 

I've managed to write some code for the potential fields. Basically it just takes an weighted average of the path to the obstacle and the object avoidance angle, and then corrects every now and then to face the obstacle. Code is on my git hub:

https://github.com/jonathanrandall/robot_cleaner_small

Let me know if you have any questions about the code.

It worked ok with this smaller robot, but one of the motors kept stalling so I kept having to give it a little push. I have five sonars, but I'm only using the middle three. I'm wondering if its the weight of the battery that is causing the motor to stall of if its just a lousy cheap motor. 

I'll test it on my DB1 robot and see how it goes. 

image

   
byron reacted
ReplyQuote
(@jonnyr)
Member
Joined: 4 years ago
Posts: 35
 

Video link is here. Right motor stalls at end and causes robot to spin around.


   
ReplyQuote
byron
(@byron)
No Title
Joined: 5 years ago
Posts: 1163
 

@jonnyr

Thanks for that code, I had a glance at it and it looks to be the business. 👍  

Right now, due to some home improvements that have take far far longer than anticipated, I wont be getting to play with any robotics endeavours for a while.  But I will be following the progress of these bot builds with great interest in my tea breaks.  (rather long tea breaks as appropriate to my age 😎  when I easily get distracted by emails and robot builds etc. 🤭 )

All your endeavours will be sure to come in very handy once I get back to my robot bench.


   
Jonnyr reacted
ReplyQuote
(@jonnyr)
Member
Joined: 4 years ago
Posts: 35
 

@byron 

You're welcome. I know what you mean. I have a million and one projects, plus kids and work. I think the key is to be consistent, even if you're only working on something for about 20 minutes a day, or just watching one youtube clip a day, or something. I've been working on DB1 for over a year now, and it's been really part time, but I can really feel the progress. I really like working on projects. IMHO there's nothing better than making something good with your own two hands.

Best of luck with the home improvements.


   
ReplyQuote
Melbul
(@melbul)
Member
Joined: 4 years ago
Posts: 43
 

Lets hope we get some more input from the OP!


   
ReplyQuote
Melbul
(@melbul)
Member
Joined: 4 years ago
Posts: 43
 

My Attempt at BB1's base programming, so far it drives the robot forwards with 5 US sensors checking for obstacles and stops if any sensor reads lower than the minimum safe distance.

lots of time and brain scratching to go yet!!

 

mel.

 


   
ReplyQuote
THRandell
(@thrandell)
Brain Donor
Joined: 3 years ago
Posts: 242
 

Hi Mel,

It looks like you have a good start on your robot’s code.  I like how you calculate the magnitude of the distance and use that to control your forward speed. The closer to an obstacle the slower the speed.  Nice. 

What you need now is a way to steer it!  Since you have wheel encoders you'll have an easier time of it.

Tom

To err is human.
To really foul up, use a computer.


   
ReplyQuote
Melbul
(@melbul)
Member
Joined: 4 years ago
Posts: 43
 

It’s a good start! Now need to add PID to the control to help smooth it out 👍😊


   
THRandell reacted
ReplyQuote
THRandell
(@thrandell)
Brain Donor
Joined: 3 years ago
Posts: 242
 

Nice work!  You've got that anti-canyon working.  It doesn't seem to have any problem swinging the rear wheels around to make a turn.

My latest robot also tries to roll over the dog's dish.  So far it just pushes them around until its IR proximity sensors pick up the wall then it turns. LOL.

To err is human.
To really foul up, use a computer.


   
ReplyQuote
Page 10 / 10