You mentioned that "There are lots of variables, even the physical angles of the sensors can change things dramatically." I was going to reconfigure my DB1 this weekend and add the US sensors across the front. Do you have the the proper angles for each sensor or will that be covered in an upcoming video?
So you need to get on the friendly side of the math if you want to make any progress here. It's just trigonometry, it really is fairly easy once you grasp the ideas.
I think this is a great statement. I could not agree more. Time invested will lead to you results. Time, basic trig, and a bit of Arduino coding abilities are the only prerequisites
I suppose we all do some simple trigonometry from time to time (like 3,4,5 for a right angle - though my fence still does not look straight 😀) but the Potential Fields paper, with all the unfamiliar symbols is a stretch too far for my current understanding.
I found a book called Trigonometry for Dummies that I may get round to reading depending on how this project progresses. I often find a deep understanding is not always necessary. In the provisional coding I have done for my bot that translates two GPS coordinates (lat & lon) into a compass bearing I found appropriate formulas to use. Not that I really understood the formula's, but I applied then firstly into a spreadsheet to see if they worked and from there into code. They work 😀, which is currently good enough for me, though don't as me what the maths COS does. Should I be understanding this, well I dont know, but if I get to read the Dummies book I may get to be dangerous.
I will talk about it soon, but the answer is that the angles and how much weight you give the sensor should correlate. I usually have one dead ahead and then orient them out at 15-20 degree increments.
In the last one I have 5 roughly at -65, -20, 0, 20, 65
The weights at those angles look like this:
const float CENTER_WEIGHT = .9F;
const float MID_WEIGHT = 0.5F; (-20, 20)
const float WIDE_WEIGHT = 0.35F; (-65, 65)
/Brian
Are the sensors the HC-SR04 sensors? Just want to confirm.
SteveG
Robotics is definitely a journey I would like to join you on. I liked your answers to the questions/comments above. I have so many projects that I want to do this year. Already the end of Feb and I still have not completed one so I am already behind. But I am the master of digging deep into rabbit holes because this hobby interconnects so many technologies that you have to know about. We will see ...
I will talk about it soon, but the answer is that the angles and how much weight you give the sensor should correlate. I usually have one dead ahead and then orient them out at 15-20 degree increments.
In the last one I have 5 roughly at -65, -20, 0, 20, 65
The weights at those angles look like this:
const float CENTER_WEIGHT = .9F;
const float MID_WEIGHT = 0.5F; (-20, 20)
const float WIDE_WEIGHT = 0.35F; (-65, 65)
Brian, is the distance between sensors critical?
It is just important that some thought is put into spacing out the firing pattern so they are not effecting each other. I actually rotate through the 5 only using 1 at a time then every 30ms or so do the next one.
Other than that, you want to cover the area in front of the robot at thoroughly as possible.
/Brian
Small update, all hardware build for the future, none of it is functional yet. I am waiting on some connections I ordered and a battery solution for the Jetson TX2 (will take 12 - 19v but from what I am reading might be erratic on 12v so I decided to get one of those 19v power banks for keeping laptop charged. I can apparently use it for 5v as well, will see how this goes.
I am also researching ROS on the TX2 and what aspects of cuda / GPUs I can utilize, Trying to determine the best way to use it. I am also looking into 3D / depth cameras, so lots to do before any of this works. I could hook up the lidar to the Arduino at this point, but I am waiting to see what I decide for a SLAM before I do. So just extra weight for him to cart around for now. But the good news is he works just as well as before for potential fields, no changes there.
And don't worry I am not getting distracted, my class is working on the Potential fields problems so that is where the videos will be staying focused. Timeline by the end of the semester (a little later this year in late may due to covid). I might through some for the higher level stuff in if i have time.
/Brian
Any chance you could switch to the Jetson Nano? It's a lower price than the TX2 and there seems to be (at least to me) more videos on the nano. Thanks.
Any chance you could switch to the Jetson Nano? It's a lower price than the TX2 and there seems to be (at least to me) more videos on the nano. Thanks.
Maybe.. not sure yet. The Jetson TX2 is faster and I already own it. I am going to start there and see what I end up doing with it first. If what I am using it for is possible on the nano and there are significant differences that it would hard to follow along otherwise with, then I might if and when I get to the point where I am making a video on that part. But since they are just computers with GPUs which I believe use the same environments (CUDA, etc), I doubt it will make a difference (unless of course the nano is not capable of running the algorithm because it is too taxing). I also mentioned that if space is a concern as I grow the project and again if the nano is capable enough I might also switch for that reason.
/Brian
My you are moving ahead fast! I guess having a teaching schedule helps. Regarding utube videos I personally don't find them the best way to review a tutorial. You might get the general gist of the subject matter but after that a written version as with Bill's videos becomes the reference point later. I once put that to another utube guy letting him know his web page was useful but not so much his long winded videos. He made the point he gets paid when people watch his videos but not when they access his web page.
I'm hooked on your videos I will definitely be watching them I think they are very well done.
There are 2 suggestions I'd like to make. First in your video code in the IDE it would be nice if you increase the font as much as possible without having to wrap lines makes it far easier for me to read the screen with the larger print if possible.
The other suggestion I would make is that when you set up your play lists for your lessons it would be nice if you had bike part one part 2 part 3 of specific things you were doing so easier to find exactly which video I need to look at and I do appreciate you putting the timestamps at the bottom which is very nice.
Unfortunately I'm currently not at home where I can actually work on the physical robots so right now I've only been watching your videos not sure when I will be home it will not be anytime soon.
DroneBot Workshop Robotics Engineer
James