Well, it is just a beginning but ... wow ... what a beginning
Hi,
as I know how problematic it is for mobile robots to move in space according to ray shooting, I was a bit afraid it will not work so good for the race.
But the results look promising so far!
Let me introduce first dummy "I do not know anything about track, lets trust sensors" example.
I do not calculate any ray-triangle intersections, I do simple ZBuffer rayshoot query.
The "racer" has just front and left/right sensors.
Hope you will like it, keep in mind this is just beginning. I know the vehicle moves like drunk
Petr
UPDATED ATTACHEMENT
Learn 3D graphics with ThinBASIC, learn TBGL!
Windows 10 64bit - Intel Core i5-3350P @ 3.1GHz - 16 GB RAM - NVIDIA GeForce GTX 1050 Ti 4GB
Well, it is just a beginning but ... wow ... what a beginning
www.thinbasic.com | www.thinbasic.com/community/ | help.thinbasic.com
Windows 10 Pro for Workstations 64bit - 32 GB - Intel(R) Xeon(R) W-10855M CPU @ 2.80GHz - NVIDIA Quadro RTX 3000
That proofs that my idea works. Great!
Your idea was very good Mike, it can save us some work.
It is simply great to know we can throw the object to any environment and it is able to navigate without need for any details.
Petr
Learn 3D graphics with ThinBASIC, learn TBGL!
Windows 10 64bit - Intel Core i5-3350P @ 3.1GHz - 16 GB RAM - NVIDIA GeForce GTX 1050 Ti 4GB
The idea can open lot of new possibilities.
And maybe the idea of rays can be encapsulated in entity system directly as native feature ( attention ... Petr will kill you ! :-X )
www.thinbasic.com | www.thinbasic.com/community/ | help.thinbasic.com
Windows 10 Pro for Workstations 64bit - 32 GB - Intel(R) Xeon(R) W-10855M CPU @ 2.80GHz - NVIDIA Quadro RTX 3000
No,
I do not kill people ... usually ;D
That reminds me ... did you sent me that Beretta already Eros?
Petr
Learn 3D graphics with ThinBASIC, learn TBGL!
Windows 10 64bit - Intel Core i5-3350P @ 3.1GHz - 16 GB RAM - NVIDIA GeForce GTX 1050 Ti 4GB
Yyyyeeeees ... that remember me something !
Attention guy, I'm Italian
;D
www.thinbasic.com | www.thinbasic.com/community/ | help.thinbasic.com
Windows 10 Pro for Workstations 64bit - 32 GB - Intel(R) Xeon(R) W-10855M CPU @ 2.80GHz - NVIDIA Quadro RTX 3000
That is cool Petr.
I think it could be easier and faster if you just have it balance the distance from left and right sensors and have it move forward.
In the race, when the front sensor senses another vehicle for instance, it could then use more sensor input to determine which side to pass on or if not enough room to brake or slow down.
Acer Notebook: Win 10 Home 64 Bit, Core i7-4702MQ @ 2.2Ghz, 12 GB RAM, nVidia GTX 760M and Intel HD 4600
Raspberry Pi 3: Raspbian OS use for Home Samba Server and Test HTTP Server
I was having a problem with the car driving through the first turn, and driving through the one on the bottom. (Your distance code is also causing unnatural "hopping", which is why I multiplied it by 10. It gets progressively slower as it approaches a target, then zooms full speed the further the target is away. EG... 40MPH, 30MPH, 20MPH, 10MPH, 5MPH, 1MPH, 0.5MPH, 0.1MPH, {Target hit}, 100MPH, 90MPH, 80MPH...)
The problem may be that you have two "Continuity" values. (Time)... Game-frame = time and distance-per-frame = time. (Essentially, your SPEED is powered by 2 times... thus, as it nears, is progressively slower, not linear, as the use for Game-frame is intended.)
You will still need waypoints, or it will not know where to go, or if it is headed in the correct direction. (Humans use them, why would a program not use them? The lines on the road, posted signs, and knowledge of the destination are our track-points.)
If anything, feelers and a track-point system as one, might maximize the operation.
(Drop the ship outside the track... It NEEDS way-points.)
The points have to be there to determine if we, or the AI, are stuck or are finishing without cheating. (Driving across the field, taking a shortcut, or after being flipped into a shorter destination, like through an overpass.)
The feelers would be great from a "Thinking" perspective, because an AI could "Ride the wall" to cut time, but not hit the wall, and be slowed-down... while being able to detect that it needs to pass someone else riding the wall. But, I do not see the feelers alone, operating the AI to win a race.
Not sure if you go here... but I find a lot of the info real helpful for my simulation programs. (Not the specific page, but the site in general.) LOL, I forgot to put the link!
Here it is...
http://www.gamedev.net/reference/list.asp?categoryid=28
I don't entirely know how these "probes" work... but is it based on this concept?
http://www.gamasutra.com/features/20.../lander_02.htm
I failed horribly at attempting to modify any of the numbers in the formula. (Not sure why you have speed resetting to 0.)
I am following this (The method), still, with strong interest. (I learn from all your code.)
Hi Kent,
you are right about the balance, that would be better.
Jason,
as I said, this is just concept demo, not final AI or anything. I just needed to test if sensors are way to go, not provide perfect AI at this moment. After reading your post I want to hang myself
What did you multiplied by 10? Could you provide code?(Your distance code is also causing unnatural "hopping", which is why I multiplied it by 10
How do you know what was intentedThe problem may be that you have two "Continuity" values. (Time)... Game-frame = time and distance-per-frame = time. (Essentially, your SPEED is powered by 2 times... thus, as it nears, is progressively slower, not linear, as the use for Game-frame is intended.)
If I shift anything at 10 / FrameRate, it is basically moving at 10 m/s.
Why I would do that? I like on this approach the track designer is not obligated to set waypoints.(Drop the ship outside the track... It NEEDS way-points.)
I agree GameDev is good resource.
Petr
Learn 3D graphics with ThinBASIC, learn TBGL!
Windows 10 64bit - Intel Core i5-3350P @ 3.1GHz - 16 GB RAM - NVIDIA GeForce GTX 1050 Ti 4GB
Bookmarks