re: absolute positioning sensors
https://store.qorvo.com/products/detail/dwm1000-qorvo/681946/
2 Likes
Something like thisâŚ.
The pole holding the cones could be horizontal though.
I think the cone laying mechanism will be the easy part, relatively speaking.
If repeatable track layouts is the goal, youâre going to need centimeter-level positioning. Localizing the bot is going to be the hardest problem for you to solve, especially without resorting to an outside-in approach (lighthouses or similar). Highly recommend you start with a minimum viable product to demonstrate that your locomotion and localization method can accurately keep track of itâs position while laying out an entire track without any appreciable drift.
3 Likes
I have a Micro Racebox Maybe that I could use somehow in coupling with the Bot as I think there is some sort of track creation feature on the Racebox app. maybe it wont help at all, Ive personally never used it & donât know how it works exactly . But yeah I think if it has a predetermined start point it should be easy.
The thing that starts getting tricky is when when maybe people are parked and in the way of the track layout. thats when on the fly track rerouting comes into play.
Maybe just making it remote control is best? Idk all up in the Air.
Maybe Driving the Bot Manually to learn the area initially is something id have to do in order to map tracks out.
Idk i gotta do a lot of research I want to make this as simple as possible for others.
2 Likes
There is a robotics competition where you have to place as many cones as possible and then pick them up. You can get a lot of ideas
For position tracking, i want to try this module for my projects https://www.ardusimple.com/product/simplertk2b/
5 Likes
Alot of riders tend to absolutely total cones in track sessions during intense duels, but I really like your idea maybe possible if they can act as Snakes.
Like 2-5 cone Segments that just Snake into Formation.
1 Like
ideas are forming⌠Many many ideas⌠thank you
2 Likes
This (automatically setting up cones) I think is a fantastic robotics challenge thatâll highlight why robots are so difficult to implement in the real world.
Itâll be easy with perfect locomotion, sure; But acheiving perfect locomotion isnât possible in the real world.
When your code tells to move the robot 2m forward, it might end up moving 2.05m forward. Turn 90 degrees left, only get 89 degrees. Gust of wind? Slipery asphalt? Relying on wheel odometry alone will lead to errors like this compounding over the whole track layout, leading to inconsistencies in the track layout. I think youâll be hard pressed to get the accuracy you need from GPS.
But Iâd love to be proven wrong!
Yeah, that too. Thatâs a whole other problem. Iâd say that building a track laying robot is ~bachelors degree-level robotics, but writing software to do automatic route re-planning is easily PHD-level stuff.
1 Like
Yeah & a literal Chimp is dreaming of it LOL
I Believe it Can Be simple.
It Doesnât really have to be perfect at first
but yes this is hyper involved and I barely have a Special ED Highschool Diploma 
But thats besides the point I think it might be worth developing that is up until all of us in the world have access to local tracks we can practice at, But Its Unlikely tracks are just going to conjure up and just so happen to be For PEVs/and or Tailored for ESK8.
Maybe in the future we will gain Access to tracks who knows.
Yes thats all Very True. I dealt With a lot of those issues in the Robotics Class I took at Pasadena City College when programming and it was a absolute headache. Theres just gotta be a way to simplify it.
If nobody Else Will I will eventually just trying to push things along
Challenge Accepted 
1 Like
We can use Google Earth to mark out a track with the measure distance feature and use the thing where it shows you where you are to walk along the track with cones.
I know it might not be very accurate but the repeatability should be reasonable and I think you can share it with anyone anywhere.
Again, I wouldnât rely on the accuracy too much and it is way less convenient than a robot, but itâs a much easier option.
2 Likes
we regularly do this in SD when a race (EE, esk8con, etc) releases a track layout. Using parking spaces as references usually works pretty good, we can throw down a pretty decent replica in about 20 minutes
1 Like
I have not seen or used this feature but it sounds cool!!
1 Like
Thatâs just a line follower. Iâm surprised thatâs even a competition (engage snark)
I would definitely start with @dimos15 's module as the baseline for everything you want to do. As others have suggested, solve the hard problem first - location location location. Dropping cones is easy. Retrieving cones is probably really hard. Youâll need vision for that. And if the cones are knocked over wellâŚ
One of the local parks has this massive paved lot
and I have a ton of cones in my carport, so I might just tag along/help out on this project 
4 Likes
It sounds like the roomba model might be the wrong paradigm for this. Having a lilâ guy scoot around a parking lot autonomously introduces positioning and cone dropping accuracy problems, when humans already know how to measure things and hold a stack of cone shaped objects. The actual problem to be solved here is the time it takes to perform the measurements.
Itâs like building a CNC router to mill out a bar of a certain length hands-free, when you really just need to buy bar stock and get a human-operated auto fence
Imagine a 4-wire, 2D version of this bot:
Find the 4 corners of your track area (the cable reels could be strung together into a rectangle), have people hold down the reels at each corner, and let the robot run through the list of coordinates. Then dispense a cone, make a chalk mark, or have a human follow behind with cones.
Better yet, just use a set of two line lasers on gimbals and tripods. Each set of coordinates rotate the gimbals such that the laser lines intersect on the ground where the cone needs to go. Human puts a cone down and walks along.
5 Likes
Very true, great insight
3 Likes
Found a cone bot!
It looks like they are using a raspberry pi 3 with an ublox gps
Here is the article
And here is the paper
https://arxiv.org/pdf/2104.14103
Ok this is epic
4 Likes
could build a big plotter.
2 Likes
Could be an option for the chassis/platform:
AFAIK, Ardupilot software can do route planning:
And you probably already know, but for centimeter accuracy, you will
need a base station to stream correction data to the rover.
Or you can buy a subscription for correction data.
Or you can get correction data over L-band built right into one gps rtk receiver;
(I own one of these)
Probably cheaper and better to just run your own base station.
2 Likes