0

I want to apply computer vision on an autonomous robot-car to position itself by detecting light sources (LED) on the field. I have an Altera FPGA and a Raspberry PI. Both will be attached with a camera. After researching on FPGA I understand that it's faster and less power consuming than micro-controllers IF you program the hardware correct and effective. This might be a problem since I have little knowledge of electrical engineering and am actually a software engineer. My question here is if it's smart to use the FPGA or just use a Raspberry PI to detect the light source and if the usage of a FPGA really delivers a 'better system'.

HieiFCB
  • 11
  • 5

1 Answers1

2

In this case, the Raspberry Pi is the better choice for a few reasons:

  1. You're a software engineer. FPGA development is entirely different from programming (different architecture, different paradigms, different language). If you've never worked with an HDL before and have no digital logic experience, the learning curve can be very steep.

  2. Computer vision requires lots of high-level, sequential operations. The parallel operations of CV are easily done with the FPGA, but you likely need a CPU as well, so you would either have to use a soft CPU or connect to an external processor. Getting the whole system to play nice may take a while.

Computer vision can certainly be done with both platforms, but from what you've said, the Raspberry Pi is a more appropriate platform. Note that it may not be the best platform (may not have quite enough "oomph"); consider something like the Nvidia Jetson development boards, which is pretty much made for this sort of thing.

uint128_t
  • 335
  • 2
  • 10
  • By analyzing the situation, I can say that using a Raspberry Pi will not fulfill my needs. I want to use a light source which blinks 25 frames per second. Therefore the device must react in 1/25=0.04ms and my this is only possible with my FPGA and not my RPI. – HieiFCB Feb 18 '16 at 15:32