• I have an upcoming project where I need to track about 8 objects in a pedestrian environment. Each object should be separated and would be moving around in the space. I will need to know their position on the X axis only but the pedestrian traffic should not interfere.

    Any idea what would be the best sensor/transmitter-receiver to use?
    Thank you
  • Izzy Guru

    Eyes ++ should do it. You will need to play around with the threshold and consider the difference in ambient lighting. A camera above will work beat as objects/people will not overlap (unless they stand on each others head!) Ideally you will need to mask off the traffic/road area so the camera just sees the pathway. Hope this helps.

  • I have used eyes ++ in the past but was not able to give an object a specific address and the result was that the objects identity got mixed. Blob 1 mixes with blob 2 etc...As for the threshold, I think you relate to the objects being in a different color or brightness. The issue is that the installation will have many pedestrians and i can not control what cloths they wear. Ideally I would have some kind of a transmitter in each object and a receiver to give me their location. That way each transmitter will have it's own id and would be independent.

    Does that make sense? The installation need to be self standing and with eyes ++ I think we will have many conflicts.

  • Yes, go and try Skulpture suggestion.

    But I suspect you have very difficult to hold the tracking on one moving object among others. I have not succeeded really even with much simpler situation. Keep in mind that Eyes(also++) will track relative to the center point, so that by itself creates a lot of unwanted ”jumping” in tracking values. Also it is hard for software to distinguish between object and shadow and pretty much impossible in changing light conditions.
    If your object is put to street by you there would be good idea to attach some kind of sensor on it that nobody else has(special color cap, IR light....).
    I would be very curious about your results. If Eyes++ and camera view really works I would be very happy.
    BTW London survey cameras supposed to be able to track a moving person even from one camera view to next and so on. So there is a technology somewhere.

  • Ahh, since you have experienced some difficulties similar to mine I would suggest to look for other options for tracking even if you use Isadora for the rest. Probably on windows you find more advanced tools. I used EyesWeb some 10 years ago with much better tracking success than Isadora today. This software is developed still so it is worth to look into.

    Also Wii could be hacked to use as transmitter or a bluetooth device.

  • Could you be more specific about what the object are, and how they will be moving about ?

  • The objects are going to be baskets that I will be able to place any kind of sensor inside. The audience will be moving the along a pat that is 6FT wide and 30ft long. I will need to know where the objects are along the 30ft. It needs to be within 1 ft accuracy.

  • I would be tempted to put infra red leds in each basket and then use a ir camera to locate them.

  • The space has lots of natural lighting that changes constantly.

    Do you think it could work?

  • I don't think that will work despite the natural light the people moving the baskets will block them when they move them won't they?

  • Will probably depend how many people are milling around them, but natural light does make it less likely.

    Its best if you describe in detail what the scenario is - makes so much difference to what may or may not work. 

  • natural light has loads of IR, i would expect problems.

    read a little about IR led tracking in the old forum:
    I'd say sculptures suggestion, maybe with different colored pingpongballs, chroma key and the eyes actor

  • Yeh. IR is out of the picture. The problem with the color detection is that pedestrians will come with various cloths in different colors and that will mess it up.

    I am looking for some transmitter/receiver that has the ability to measure distance or proximity of a specific object.

  • As for the environment. It will be a 6'X30' bridge that people will be moving back and forth on. They will carry the baskets (where i plan to put the transmitters). There can be 30 people at a time there and each one can move any basket (up to 8 baskets). By knowing which basket is where, I will manipulate localized speakers.

    I was thinking also of something in the RFID family but the cheaper and simpler the better.
    Arduino can also be a solution but I am still a bit lost.
  • Beta Tester

    I would consider exploring the sorts of RF tracking systems used in products like TTA StageTracker.

  • It sounds very interesting and would love to test it, if this system is affordable.

    Thank you
  • Beta Tester

    I suspect that system is not affordable and, to the best of my understanding, it won't perform the functions you would need anyway as it does not actually output tracking data but I figured that you might be able to look into sourcing or building a system that operates on similar principals as that system appears to meet most of your technical needs in terms of tracking multiple targets without consistent line of sight over a relatively large area.

  • Most rfid people are familiar with is passive. Range between reader and tags is up to about 3 inches

    You would need to use active rfid (the tag has its own power supply) to get the kind of range you need with rfid. There are various companies that do RTLS (real time location systems -using those kind of technologies to find position as well as identity). They often work on just locating which room something is in, but talk of doing positioning to 1ft accuracy. Suspect it will not be cheap.
    Quick google search throws up this page (not saying this will do what you need, but is moving towards it)

  • RFID-arduino-serial-isadora is technically possible. You will however have to have many receivers, because there is no (easy) way of measuring signal strength, only presence. So no triangulation.

    GPS maybe? If its accurate enough. Plenty of arduino GPS resources out there

  • I think that GPS is not accurate enough. I believe 30ft range.

    Today I will be contacting some Active RFID companies and will update.