sotolab-robotic-prey-fish

First time setup and running:

• plug in camera and transmit board
• place robot near the start position and turn on (all the way, both clicks)
• open up controller.py
• find ports:
  • find transmit board serial port by opening at terminal and typing ls /dev/tty.usb*
  • in main, switch out port_t with the listed port
  • port_c is the camera port and is either 0 or 1. find with guess and check
• if the camera or tank have been moved:
  • find the new bounds of the tank from the camera's view with calibrate_setup.py
  • copy those new bounds into main
• click the ol' "run python file" button and watch the robot go

Code overview:

cyberbot directory

has code for micro:bits, generally this does not need updating

• to upload new code, paste into main.py, open just the "for running" directory and upload from there
• receiver.py and transmitter.py store the code for those chips
• if stuck, see https://www.youtube.com/watch?v=eSGJLu1kqyg
• display.scroll("text") is useful for quick debugging

src directory

has code for running the bot

• controller.py is the top-level class that creates instances of data_handler and video_processor
• paths.py and point_tracker.py are also necessary helper files
• analyze_data is for post-processing, and kalman.py is a kalman filter

testing sub-directory

has some test files that may be useful or may be antiquated/broken

• calibrate_setup.py is for finding the bounds of the tank from the camera's view (in pixels)
• collectImages.py is for taking photos from the video stream
• drawTrajectory.py draws a path
• imageProcessing tests the camera subsystem
• pathTrackerSim tests the path tracking performance, but likely is unrealistic and not useful
• sendCommandsSimple tests the robot subsystem
• wallPotentialField visualizes the artificial potential field algorithm

data directory

should ideally take care of itself. the DataHandler object controller.py instantiates should populate it with data files as directed, with sub-directories and filenames corresponding to the day's date and time

Known issues:

• calibrate_setup.py has problems
• saving video files is untested but likely works
• pixels to meters conversion should be automated
• new code does not implement desired angle tracking (gain for beta = 0)
• new code does not implement delayed start functionality (robot starts as soon as the file is run. it could be useful to implement a delay that waits for the user to press a key for the trajectory to begin. this is partially implemented with the "go" boolean in the VideoProcesor)
• kalman filter should implement an actual motion model (prediction step)
• radio communication may be dropping some transmissions