Notifications
Clear all
Topic starter
2020-01-17 12:27 am
Positioning systems may be rather an ambitious topic as my brief research has only thrown up expensive components. But maybe there is a less expensive way? As we know when moving a robot via motorised wheels the small differences in each motor often means the robot is biased in one direction. This is easily compensated for by taking into account encoder readings to speed up and/or reduce the speed of the individual motors appropriately. This does work well but with limitations of wheel slippage. Indoors this is not so much of a problem, though it can be when the robot is 'bounced' by going over a door threshold or once wheel on a carpet and another on a shiny tiled surface. Outdoor robots suffer a lot more. Bouncing about over tufts of grass or gravel surfaces etc will cause havoc if just using encoder feedback. Including the use of a compass bearing chip can help, but ultimately the ability to know exactly where the Robot is and to feed this back into the control loop seems to be the only way to go for ultimate control. I wonder if you have any thoughts on this, and if this would make a good subject for one of you videos, even if it proves somewhat a strain on my wallet. (and yours too ? but DB1 would surely like it) - Edit maybe this could be one execption where you do do a sponsored episode.
2020-01-17 2:19 am
But maybe there is a less expensive way?
Using an IMU (Inertial Measurment Unit) is an inexpensive way to maintain a consistent relative orientation. It won't provide absolute positioning. But it can correct for having been thrown of course, or sensing whether a 90 degree spin actually resulted in turning the robot body 90 degrees.
The subject is so deep it requires a multiple video course. Paul McWhorter has a very good course on how to use IMUs.
Paul is up to 23 videos on this already and still going strong.
About all Bill could do in a single episode is basically explain what an IMU does.
In fact, Bill already has touched on the IMU with this video on the MPU-6050
DroneBot Workshop Robotics Engineer
James
2020-01-17 2:23 am
From what I read about DB1 is it will be able to navigate using LIDAR and use ROS code.
My approach is using webcams to find recognizable "landmarks" to identify the robot's location and position within a house.
Topic starter
2020-01-17 7:07 am
Thanks for the link to the 9 axis course I will be following up on that. I actually have a 9 axis imu and have had a brief initial play but have yet to succeed to get my outdoor robot get to an accurate point once it has gone 50 meters. But this was just an initial play and I will hopefully get better results in the future. A cold wet wintertime field with overlong grass does not encourage more experimentation at this time.
@casey
I've also got a couple of LIDAR units to play with but for the 360 degree one, on a field with not too many landmarks, I think it will be relegated to detecting unexpected obstacles rather than assisting in navigation. I got a pixy2 camera with the hope of finding landmarks especially to assist in precise positioning when approaching a base station. I'm not sure if I want to put up a lot of landmarks in my field just to assist the foxbot navigation but this is on my list of possibilities. I will be following any information you may share on the forum as to your success with this.
I shall certainly be going further with the options you both mention but I was also hoping @dronebot-workshop might present what could be achieved with a positioning system to weigh up all the possible navigation aids, though probably I (and Bill) will find this route is too expensive
2020-01-17 7:43 am
So this is your robot?
https://forum.dronebotworkshop.com/user-robot-projects/foxbot-a-bunny-chasing-robot/#post-1216
I would have thought gps would be the best solution outside. Without seeing your "field" it is hard to imagine what obstacles you have. Are you serious about the rabbits? If you want to protect a home vege patch surely a rabbit proof fence would be the easiest solution?
2020-01-17 8:15 am
@casey
I would have thought gps would be the best solution outside.
GPS is overkill, how about a sextant and a John Harrison clock! ?
Now, that is an interesting thought, building a copy of H4, with a 3D printer! ?
Topic starter
2020-01-17 9:22 am
@casey
Yes that my robot. My experiments with 2 gps boards floundered a bit when, despite two identical gps boards and antennas, one board consistently locked on to fewer satellites than the other. I've yet to see if this is just an antenna problem. Normal GPS is only accurate to several metres and not good enough for my purposes, but differential GPS can get to centimetre accurate but at a cost. Maybe a cost I have to pay. If you want to imaging my field, think of a level paddock surrounded by hedges. You may also like to image a peacefully grazing horse with a straw hat, but in reality just grass with a few holes where bunnies have dug for roots along with their fertiliser deposits. Put up a fence! Don't be a spoilsport. Bunny fencing must be dug in 6 inches below ground or they tunnel under and the field has two field gates where this is not possible. And I've no wish to have a dog or to shoot them.
A sextant or even a star gazing camera may be possible for those cloudless nights. Maybe a moon gazing foxbot could howl on full moon nights, that should scare the fur of the bunnies tail and give rise to some local werewolf rummers.
Topic starter
2020-01-19 4:48 pm
My mission - to enginer foxbot to boldly go where no robot has gone before. ?