Replies

  • Do NOT forget for obstacle avoidance using the Xbox Kinect requires a full USB2 bandwidth (USB2 requires a fairly new PC not just a microcontroller). Mini-ITX boards (17cm x 17cm), Nano-ITX (12cm x 12cm) and Pico-ITX (10cm x 7.2cm) have USB2 so you would need to accomodate them on to your vehicle. With a fully fledged PC running though you can get more user friendly programming languages involved too.
  • I JUST bought APM and all products with it except the wireless part since the xbee adapter is out of stock
  • Shall I use Ublox or Mediatek as GPS
  • ArduPilotMega is for flying platforms with aileron control (twisting along axis), elevator control (up/down), throttle control (power) and rudder control (left/right). You want throttle control (channel3) and rudder control (channel4) for a UGV as the other two are completely useless to you.

    Now the problem is that the APM coding makes travelling as direct as possible (as the skies don't have bollards and cars). Your UGV will have more than likely several things to deal with on its journey so you need some detection and avoidance hardware and software.

    Detection and avoidance systems are ranging sensors with software to take the sensor data and create a map of some sort so you know if you are going to crash and can take action if you are. Decent cruise speed is 30mph (13m/s)

    Infra-red sensors - fire a pulse of IR light and time its flight. The time of flight cut in half * speed of light is the distance. Sharp do sensors that do this but range is in small clumps so its not great. The furthest sensed is 150cm with about 300cm being detectable but accuracy is not great by this point. At cruise you'd have have 0.25 of a second to steer clear of something (and thats with accuracy in your favour).

    Sonar sensor - fire a pulse of ultrasonic airpressure and time its flight. Time of flight cut in half * speed of sound is the distance. Maxbotix do sensors the do good range for this with good accuracy however while moving this becomes problematic. I shall come back to this later with another idea for you.

    Laser - position of laser dot in field of view gives distance for certain point, LIDAR does time of flight. LIDAR is very accurate (some down to 3mm) with a range of 6-10m but its quite complicated to get done. FoV position is only as accurate as the camera used to range it but I personally have got range of over 100ft but accuracy is in feet by then. Revo LDS from their hoover is $400 and accurate to 3mm with distance to 6m.

    Camera - recognition, stereographic etc. Cameras are heavy on processing (WAY more than an Arduino can provide) but can work very well. Xbox kinect system has proven its potency but you need the complete bandwidth of a USB2 connection 480Mbps (Arduino do 115.2kbps easily and a little higher but not even 1Mbps).

    RADAR - time of flight measurement again. Simplistic units can be made from spares apparently but I am yet to find any :S Volvo have these units in them so a crashed one could be a gold mine for fun. Accuracy is same as most of the EM units and as the frequency is so high range is significantly improved.

    Thats it for the sensor side but there is something I would consider useful. The sonar sensors by Maxbotix are great when sat stil but with wind (changing the air pressure) and movement (varying airpressure around model and weather). One of their sensors (called AE) has an analgoue envelope output which is the echo from the pulse sent in its entirity. This may not sound like much but I you were give this with an EM emitter/receiver setup it is called RADAR A-scope system. Using this and some post processing I think you could create a kick ass setup but it would require multiple readings (possibe multiple sensors) and some very heavy coding.


    Hope that helps
  • I am using RC car as start
  • Unmanned Ground Vehicle
This reply was deleted.

Activity

Jose Araujo liked Jose Araujo's profile
Aug 29
spencer harvey liked spencer harvey's profile
Jul 9
More…