Telemetry with a Boe-Bot or two?

ahay

I have a project I have been toying with for a while, and I am looking for some feedback on the feasibility of it actually working and if so, what sensors would best suit my needs?

I have a Boe-Bot(master) that will roam freely, making decisions where to go based on IR object detection and the bump sensors. Now, the hard part, I'd like a second Boe-Bot(slave), to receive location data from the master, more than likely GPS data, and follow the master's path, with the ability to vary the distance at which it follows. The maximum distance would be about 3 meters. It will also need to sense when movement of the master Boe-Bot stops, and then pull up next to it and wait for movement to begin again. Finally, the keychain transmitter would be used to start/stop the slave, and set the distance between three values, let's say close, not so close, and max(3m).

Utilizing the GPS solution, I'm assuming I would use individual GPS's on each Boe-Bot and an RF communication link between the two. The slave would compare the masters location with it's own, and follow according to the specified parameters. My questions are as follows;

Will GPS give me the accuracy needed to maintain a 2-3 foot distance without the slave running into the master?

Is there a way to accomplish the same task without GPS, maybe triangulation of the masters position using the RF signal?

Is this application too advanced for the Basic Stamp? I hope not, because I really like the product and the community surrounding Parallax!

Thanks,

Al

Paul Baker

You cannot guarentee GPS will give you that much resolution, RF triangulation would be pretty much a no go at that close distance either. My best suggestion is to use either an IR becon, or use an RF link where the master bot communicates what movement it is performing and the slave has a buffer of these commands that it replays at a delayed time (no movement by master bot puts slave bot into a pause on its replay buffer until the master starts moving again)

SPENCE

Paul.

This may be taking it to the extreme, but how would it work to hack a pair of the predicessor 's of the ping and split the operation. A send to b, b reply to a. Time delay measured using 28sx in machine language at each end.

73
spence
k4kep

ahay

Thanks Paul, that answers my initial questions. Without going into too much detail about specifics, this is really just an example of the application. Ultimately the master would not be a robot, but a person, which makes the problem even that much more difficult. Based on that, IR could potentialy work for location, and then using RF to communicate operating parameter changes. What about using a digital compass and memsic accelerometer to determine movement of the master? That would eliminate potential problems with visibility of an IR beacon, and timing delays could accomplish basic distance control within a meter or two.

Paul Baker

Hmm let me think on it a bit.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔

DukeEgr93

I am a bit confused by the eventual application - what are you assuming the "master" person will have available for the follower robot to track? Is it possible that the person could be wearing some form of transmitter/receiver pair or should the follower (which, to a degree, sounds like a well trained dog) have pattern recognition coupled with a receiver to get the "how close" information?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
____________________________________________________________
Dr. Michael R. Gustafson II (mrg@duke.edu)
Lieutenant Commander, Civil Engineer Corps, USNR

ahay

Ultimately I'd like it to track the master via GPS coordinates relayed from a device the master carries. Pattern recognition would certainly be useful, but not necessary. One of my initial thoughts was to use a sensor that is commonly found in auto focus cameras to determine distance to the subject. However, that poses the same problems as an IR beacon or pattern recognition and would require the master to always be within LOS. In the 'ideal' world, the master would have an IPaq or something for GPS location, issuing Start/Stop commands, setting the follow distance, and reviewing additional sensor data picked up by the slave which would all be done via RF.

Thanks for all of your input guys, great stuff!

DukeEgr93

If GPS really is the primary objective to follow, then you're going to have to settle for its spatial resolution - your robodog will be able to heel, but only in about a 6' radius of the master's leg :-D

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
____________________________________________________________
Dr. Michael R. Gustafson II (mrg@duke.edu)
Lieutenant Commander, Civil Engineer Corps, USNR

kelvin james

Anyway you try to do this is, you are going to have to write one huge program to make it work. At the distance you are trying to get, GPS won't happen. Even with WAAS with an unobstructed view, outside with no interference, you might be able to pull out 10 ft. If it was in that range, you could probably use 2 GPS units to talk to each other, one to set the "track", and the other to follow it. Without GPS, you have to make some sort of co-ordinate positioning system, that the follower can get the control feedback to establish its position, compared to the leaders' path. Probably the best way is to get the leader to "datalog" all the distance, heading, time, and the input from activated sensors at a given point. This info will give the follower the data to look for, and compare its movement against those values. You could use an encoder on a wheel for speed and distance, a compass module for heading, IR or sonar sensors for object detection ( which would be the basis for comparison to path direction/ data co-ordinate to look for ) and a clock/timer. With the data from the leader, simple math could be used to vary distance. But this is all in theory, as there are many factors, such as a difference in overall mechanical operation ( one moves faster than the other, slight variance in compass heading ) that could cause errors. So, a lot of code would have to be written to overcome the " errors ", as the follower is going blind and will not be able to recover from a position loss. You may be able to work it so if the leader sees that the follower gets out of a certain distance ( or a position achieved confirmation has timed out ), or outputs a position loss, it could backtrack to a point where the follower could re-establish the correct tracking. It sounds very processor intensive, a lot of stuff happening at once, so i don't think a stamp could handle it, but probably a SX would. Anyway, just spouting off here, sounds like a good idea, hope you have some success.

kelvin

Bruce Bates

Al -

There is probably a very simple method you could use to accomplish this master-slave coupling that you're trying to achieve. However, it probably violates a restriction that you haven't YET mentioned

The trailing (slave) unit could be on a leash or tether from the master, and through this tether have the directional commands (just) used by the master transmitted back to the slave. Then the slave only needs to delay the execution of the directional commands, based on the distance from the master to the slave. So long as there wasn't too much slack in the tether cable, the master could even get out of sight (around a corner) from the slave, and the slave could still catch up to the master momentarily, possibly reeling in the slack tether cable as it goes around the corner.

Just another thought.

Regards,

Bruce Bates

Doug Reiland

Is there some wireless technology where strength of signal could be used to calc. the distance between leader and follower.

ahay

Thanks for all the input. Right now I'm just kind of doing a feasibility study, and I'll post some more of my thoughts and ideas once work slows down a little to give me time for my own hobbies.

Kelvin, my original plan is to use GPS for both the master and the slave. After a lot of thinking, I could probably live with 10', as long as it is somewhat consistent and reliable. I can't have the slave wondering of somewhere and getting into trouble.

thanks again!

Paul Baker

The system error introduced by the US government is due to the time provided by the sattelites is rounded to a lower degree of precision (If I remember correctly). This time value is used to calculate the signal flight time from multiple sattelites and triangulates your location (same concept seismologists use to locate the origin of an earthquake, but in a single source/multiple receiver setup instead of GPS's multiple source/single receiver). Two GPS's can be used to obtain a more precise measurement by setting them up a known distance from each other. Now if the error is fixed (does not change) after calibration of the two units, they could be moved apart from each other and still maintain the higher degree of accuracy. However if the error is variable (changes over time) the differential GPS setup will be of no use to you, you'll have to research whether the error is fixed or variable.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔