Sensors for a self-driving toy car

Projects General Guidance
1

Hi,

We are two dutch high school seniors working on an 80 hour school project. We are making a self driving arduino based car (20 x 40 cm). We want the car to be able to drive to a beacon ~50 meter away. We want to put 3 or 4 sensors on our car. Using these sensors we want to calculate the angle to the beacon, so the car knows in which direction it has to drive.

We have researched which sensors and beacon to use and came up with different possibilities, each with its own difficulty:

Directional antennas and a radio beacon. We will need to test in which setup these antennas need to be placed. The problem with using antennas is that these are pretty big and bulky, and we are not sure if our car can handle that. Especially since it also needs two receivers.
Sound sensors. The main problem with this one is that the sound sensors we found for the arduino are fit for close range but not for 50 meters. And we are doubting if using amplifiers fixes this problem.
A flickering light source (1s on, 1s off) and photoresistors. Using the ΔΩ from the photoresistors we can calculate pretty well from which angle the light origins. We felt like this one was a bit too easy and that we could do more.

We also came up with something totally different: using a camera and and high contrast colours. We wanted to let the car stop when it encountered a red contrast and continues driving when it encounters a green contrast (like traffic lights). Because this is something totally different we decided the car approaches the contrast straight on. When we discussed this option with our teacher he said that using camera input is heavy on computers, so we weren’t sure if an arduino could handle it. We are also not sure if it’s realistic for us to use camera input, because it is very difficult.

It’s hard for us to estimate which of these possibilities are realistic and which arduino we should use. So we wanted your advice before we spend the money on parts for a project that is way too hard/impossible. All help is appreciated!

2

The camera approach indeed will require a more powerful controller (Leonardo, RasPi...).

Antennas for high RF frequencies are not really bulky, I'd think that directional optical sensors are not really smaller. See (wikipedia) "instrument landing system", for more or less sophisticated approaches.

I'd go for the flickering light source. Photo resistors may be too slow, faster photo diodes/transistors may be required, or IR remote control components. Consider the technical aspects (and safety!) of distributing light recognizable 50m away, across the entire starting area. You may choose something like a traditional light house, with a bright beam moving around 360° (or less). The car may be equipped with several wide-angle receivers around it, for raw orientation, and narrow-angle forward sensors for straight navigation towards the located beacon.

The scientific aspect of your project may be a feasibility study of the approaches, with variations of the specific installations. This can be done with a stationary sender/receiver pair, with only the final solution combined with the car. The car itself is another item, you can use a ready-made RC car or robot chassis, or build your own.

Don't forget to make you car stop before it crashes into the beacon :wink:

3

I'm also interested in self-driving model projects. Any updates on this? Bookmarking this thread.