End detection ideas?

I am working on redesigning an autonomous robot designed to navigate to a specific point, drop a load, and navigate back to the start. I have the navigation working nicely using a combination of a pair of pololu IR beacons and other sensors to detect walls, edges, etc, however I am trying to figure out a good method of knowing when I have reached my destination. Might anyone have any ideas on this subject? Thanks!

Hello,

What does the destination look like? What are the available cues that would allow a person to recognize that the robot has reached its destination?

- Jan

Well, the course itself is on a table, and the end is a point in the center of the table (left to right), 2" in from the far edge. There is a 6" wall along the far edge The second IR beacon is set up at that point in order to guide the robot in, but of course doesn’t give any distance data. Other than that, there are no real differences between that point and the rest of the course, unless I set something up.

The Sharp IR distance sensors would be great for detecting when you are 2" from a wall - just make sure to mount them a few inches inside the chassis of your robot, since they have a lower limit of 4". You can even use two to make sure that you are lined up at a right angle to the wall…

Thanks for the suggestion. Unfortunately, using the sharp sensors, I would be unable to distinguish between the end wall, and any other wall I might come into contact with. As the robot will need to navigate an obstacle course prior to reaching the end, this wouldn’t work very well. Sorry, I forgot to mention that.

Any ideas? Or other places I should ask? Anyone? Or is it just impossible to detect a specific point/obstacle, as opposed to detecting just any obstacle? Thanks :slight_smile:

Hello,

There should be a way to determine a position if you can find what is unique about the position. On the other hand, if you cannot explain the position to a person, then it would be difficult to expect your robot to find it.

For example, you could detect the walls you used in specifying (to this forum) the position. If there are many places in your robot’s world that fit that description, then you will have to narrow down the definition of the target. If there are two similar positions (at the opposite ends of the table, for instance), then you could keep some track of your robot’s position to determine which of the two locations you are at. Using sensors like encoders to determine your absolute position is not very accurate, but if you combine that with distance readings off the various sides of the robot, you should be able to get a decent idea of where you are.

- Jan

True, currently asside from the presence of the IR beacon at the position, there is nothing unique about the spot, other than the geometry (2" in, centered in this specific case). That does not, however, preclude the posibility of making it unique in some way. Determining the absolute position could work, but as you said, it is not very acurate, and acuracy is VERY important in my design.

I guess from what i am hearing, as things stand, with only the IR beacon to distinguish the end point from any other wall, it would be dificult at best to determine when I have arrived. That granted, what might be the easiest way to mark the end point so my robot can acurately detect it? Some mark on the ground, something on the wall, some measurment off of the IR beacon? Something else I am not thinking of? Pretty much anything is a possibility, as long as it is acurate. One other reason I don’t really like the absolute position tracking is that it is highly dependant on the geometry of the course. If the course changes, it would require reprograming before my robot could run it. I would prefer something where I can just set up an arbitrary point, with the IR beacon and whatever sort of marker is neccesary, and have my robot be able to run it.

Thanks for the ideas, keep them coming :smiley: !

Hello,

If you have the option of modifying the destination, that should make your task a lot easier. The tradeoff then becomes how much do you want to modify the environment vs. how easy do you want to make it for the robot. For instance, a small reflective spot or magnet could mark the spot, but that doesn’t help your robot find it.

If you can do so, it might be interesting to do something like paint a larger area around the target and make the paint gradually darker as you get toward the destination. Then, you could have a few sensors on the bottom of your robot that would give the robot the ability to detect which way the target is. (The final destination could still have a magnet or some other unique identifier.)

- Jan