I am building a project with a quadcopter. I want to be able to detect specific objects (such as keys or animals).
My current thought is using an RF transmitter, however, I would like to have those transmitters last as long as possible and as small as possible.
Does anyone know of any better ideas of how I can detect objects?
paint the objects a bright color
If your drone are big enough you can implement object tracking using OpenCV on Raspberry Pi.
If not you can send series of images by transmitter to computer on the ground, implement it there and after that send command "object identified" to quadcopter.
But response of second solution is not too much quick.
There is also possibility for best performance implement the same algorithm to an FPGA but it's not so easy.
paint the objects a bright color
One of the problems is if a person is wearing that colour, it could get awkward, lol.
If your drone are big enough you can implement object tracking using OpenCV on Raspberry Pi.
If not you can send series of images by transmitter to computer on the ground, implement it there and after that send command "object identified" to quadcopter.
But response of second solution is not too much quick.
There is also possibility for best performance implement the same algorithm to an FPGA but it's not so easy.
Its a 680 mm (motor to motor) frame, so it would be definitely big enough for the Pi. I am working on building a circuit board which will contain the sensors (MPU 9250 (i2C), LiDAR sensor (i2C), Motor PWM, and RC communication (UART)).
As it requires a few connections and 6 PWM channels (it's a hexacopter), I am looking at using MK12DX256VLK5 (the MK10DN64VFM5 doesn't have the pin count), or MKV58F1M0VLL24 Microcontroller. The MK12D only runs at 50Mhz, however, should be ok if I use the Pi as the brains, while putting the basic higher lever motor control logic (actual motor control is done via the ESC, but requires PWM to communicate), i.e level flight, turning, etc.
The MKV58 has a faster processor, however, I don't know if the algorithms would run well even with the faster processor (240 Mhz). However, may use the MKV58F1M0VLL24 for expansion reasons even if it can't handle the algorithms.