self driving TC3

[trw]

i was searching google image search for a picture and came across this....a car that looks like a TC3 fitted out with sonar to drive itself!

http://this.is/torfi/robotblog/index.php?getInfo=1&projectID=33

[AudiTT-Quattro]

Neat! I wonder if it can make course corrections at the speeds we're talking about!

[minimole]

Sorry, but im pretty sure that that is the kyosho alpha ep 2. Basically, a 4wd shaft driven TC, yeah, similar to the TC3

And yeah, i did own on, it was my first/second car(i ot two cars for the same birthday)

Duane :D :D

[CENthasizer]

well if you look at the bumper you will see its gone through some major testing!

[dog8spam]

Im working on a similar project for a tech class I have now. Im starting from the basics (touch sensors that must hit stuff) and working up to the sonar, partly because it will be more reliable and partly becuse I can afford the sensors now. Its really very intresting, I think more RC people should get into robotics.

[Chris LaPanse]

I agree. Robotics is interesting. Even though they don't go as fast as my RC stuff, it is still fun

[microrcdude]

well, i would do robotics, but i dont have much money

[dog8spam]

If I get an SX key, wich would allow me to program raw chips, I would happily distribute chips to RC guys. It might be awhile off as the SX development kit costs about $150, but a raw chip costs only about $5-6 once you have the key. Most other electronic componenets are dirt cheap (sub $1 range).

If you want to get started sooner I would suggest www.parallax.com check out the Basic Stamps, they are very easy to program, even for those who know nothing about computers (me included).

[Piggy89373]

dog8spam, instead of contact sensors, look at IR sensors. A good IR sensor and IR led would look pretty impressive to the naked eye, although it would also be slow, just not as slow as contact.

[dog8spam]

In the end I will have three sets of sensors, touch, IR, and sonar. The touch sensors will be backup for if the IR or sensors fail. I also plan to have a seperate chip that counts how many times a sensor fails and will disable it after a certain number of times or slow the truck down so it can travel at safer speeds in hostile territory. The IR sensors are from Parallax and tell when something is within 1 foot of the sensor. They will primarily be used on the corners and back of the truck. There will be two sonar detectors in front slightly angled out, each have a range of 1m and can acutally measure distance istead of just sending an I/0 like the IR. This will allow the truck to turn at different rates, slower at first and then, if neccisary, all the way over.

I cant do this all at once, each IR costing $15 and sonar about $30 so this will be an ongoing project. At first the truck will just have to go really slow, I can also spend more time tweaking the software.

Ill be using whiskers for the touch sensors, you can barely tell theyre there from a distance. Just imagine a really thin piece of wire placed on a car much like a whisker on an animal. Fix the end and run it through a washer so that it doesnt touch it. Only when something hits the wire does it hit the washer connecting a circut wich can be sent to the chip. This may be hard to picture with out a, uh, picture, but I wont have huge paddles or anything sticking out of the car, just a few wires.

[R3VoLuTiOn]

pretty cool stuff. yes it is a kyosho alpha ep2

[lerningdriver]

i was searching google image search for a picture and came across this....a car that looks like a TC3 fitted out with sonar to drive itself!

http://this.is/torfi/robotblog/index.php?getInfo=1&projectID=33

sorry to let you down, but with that size computer board on it, there's just no way it will do anything impressive. The car will never be able to drive by itself. People have made cars like that with advanced lasers and 30 different top of the line processors which weigh about 2500 lb total and STILL couldn't get it to drive itself.. so that car just aint going no-where :cool:

[trw]

lol...i wasnt excited about anything....i just thought it was interesting so i posted it.

no car can actually drive itself cause it wouldnt know where to actually go...you'd still have to have a human driver just that the car could help out a lil.

[dog8spam]

Learning Driver-
I usually never flat out deny someones post or say they are wrong, but I cant lie, you are WRONG.
Before the flaming starts let me say why...

The cars you are talking about (I am assuming you are talking about the Grand Challenge) are much more complex. The Grand Challenge was made by the military in order to jump start thier efforts to get computers to be able to recognize the physical world around us and act accordingly.
These robots like the TC3 the hobbyist are building only use simple proxy sensors to tell if its going to hit something. Theyre not going 40mph, theyre not going off road, following GPS, they avoid anything that sets of a sensor wheather it be a cloud of dust or a wall. That is why the processors can be so small. The programs are very simple, mostly a list of IF THENS
IF frontsensor=1 THEN driveback for example.

The Grand Challenge cars are much bigger, are going off road, and have much more complicated systems. These actually create wireframe models of the world around them. Proxy sensors dont cut it in the Grand Challenge, the ground isnt level, not all things it sees are solid, a hill could set of a proxy sensor. With a higher level of complexity reliablitly goes down.

No the TC3 wont be able to run a race or anything like that, but it can drive itself. By using encoders you can tell it to go a certain distance then come back, you can have it follow walls, lines, colors, light, or just wander aimlessly and avoid objects. Like I said earlier, this is a really bad example with only two sensors all itd be able to do is avoid head on collisions with something.