Partie interactive du site pédagogique ELECINF344/ELECINF381 de Télécom ParisTech (occurrence 2011).


Copterix : communication between Gumstix and STM32

Today, we have resolved the problem concerning the communication between the Gumstix and the STM32. Our initialization of the serial port wasn’t good because we didn’t set raw mode. Now we have a good communication at 1MHz between the two boards and we’re able to send around 300 packets per second (21 bytes per packet) . We are able to transmit all the data from sensors to the Gumstix with a precise protocol previously explained. We are also able to receive data from the Gumstix in order to command the motors later.

For the time being, we gather information from gyroscopes, accelerometers and magnetometers. These sensors are enough to use the Kalman filter.  Besides, this filter has a little problem when we are doing brutal movements. A first idea to resolve this problem is to optimize our filter to be faster and to take into account big sensors variations.
After having optimized the Kalman filter (for now, we just split the 3D render process and the Kalman process, and we use ZMQ to exchange data), it became more efficient.

Tomorrow, we will add a Sharp to measure our altitude because we don’t use a pressure sensor anymore. We will also begin the communication with the motors. We will have to parameter the Kalman filter in good conditions, and finally optimize it deeply before putting it on Gumstix.

Sur le même sujet :

  1. Copterix : ubuntu sur gumstix && communication STM32/Gumstix
  2. Copterix : FQA, PCB et Communication avec le STM32
  3. Copterix : Gumstix
  4. Copterix : on passe entièrement sur la Gumstix !
  5. Copterix : les aventures commencent

3 comments to Copterix : communication between Gumstix and STM32

  • François-Xavier MOREL

    Remember that in nominal usage(inside the actual copter), movements will not be that fast and the copter won’t be inclined more than 30° from the horizontal plane. Of course, making it react faster to large amplitudes is good but not at the expend of an accuracy loss around the horizontal plane (where even 1° is critical to maintain the stability)

  • Thanks for the advice !