To test this feature, visit your live site.

Driving bittle from Raspberry PI

Hi folks,

I got the raspberry pi connector on top of bittle, set up and everything works. I can send commands to from python, and Bittle mostly does them. Now, a few problems:

- the interface from Python is extremely coarse - you can give commands and some time (in seconds), but a better one would be "run x instances of the loop". Is that not available yet, or did I miss it?

- the communication is also very error-prone. I get into modes where the skills don't work, but things like turn off/balance do. How do I know the status of the channel, and is there a way to reset it without a more general reset?

Generally though, should I be investing in the serial communication, or should I start looking into i2c (and flip that switch) which I presume means driving the servos directly?

If this question is answered/portable from a different Petoi platform, I can crib from there, just point me to it please - couldn't find it on my own.

Thanks,

Bear.

PS: posting to the forum seems broken, the "select your post category" can't select any option, so "publish" is grayed out.

3 answers4 replies

7 Comments

Amit Banerjee

Jul 18, 2021

This looks like a very cool project. Do you have instructions somewhere?

Knut Bear

Jul 20, 2021

Replying to

If you give me a bit I'll upload a newer version that's a bit cleaner/supports more things. You can also set up web streaming so you can make a little remote controlled camera.

You wanna roughly do the following:

get your bittle working/calibrated with the regular serial module. should be in the instructions/videos that come with the kit.
solder in the raspberry pi connector. I'm using a zero w and it's enough horsepower
get a camera but note if you're using pi zero you need the special cable that's smaller at one end.
on raspberry pi, install git (you can find instructions)
clone the opencat repository
drop the code above in the serialMaster directory
install some video streaming - I found this works pretty well
keep a terminal open to the raspberry pi and run the python program that reads the keys
watch it on the browser

There's a slight delay (under a sec) but I navigated around the couch from another room, etc. Pretty neat. Not at all stealth.

You can also buy a battery holder and some beefy batteries that will last you hours, but it's a bit more involved project to add a connector - I found that bittle is powerful enough to lug them around, and you can velcro/rubber band them to the underside in place of the original battery, easily.

The steps above assume a lot of stuff - like you know how to set a raspberry pi, etc. so depending on your knowledge you'll have to chase down a bunch of other instructions, let me know if you need more help.

Knut Bear

Jul 11, 2021

I think found my problem. in OpenCat.ino, line 632 you have:

if (strcmp(newCmd, "") && strcmp(newCmd, lastCmd) ) {

but the treatment is inconsistent for skills. Balance, specifically, writes something else in lastCmd even though it's a skill. Check out line 711:

strcpy(lastCmd, "balance");

So that strips the "k", meaning you can send "kbalance" in a row, whereas if you send other skills like crawl forward, etc. it gets ignored. Makes sense I guess if you're on a remote that keeps sending the same command over and over again, you don't want to restart the sequence.

Somehow - haven't finished tracing all the code (like around line 650), if you have a sequence over serial like "kcrF" -> "kbalance" -> "kcrF", the second "kcrF" will get eaten.

Would be nice to document the state machine explicitly, clean it up a bit and break up the code into more functions - but I can't complain, it works and wouldn't have wrote it myself, so here we are.

Rongzhong Li

Jul 12, 2021

Replying to

Yes, I realize there's some bug in the parsing and processing between skills. Sometimes a skill will be ignored and can only be called by press another key first. I'm getting dumb recently and haven't cleared my mind to restructure the logic.

Knut Bear

Jul 14, 2021

Replying to

@Rongzhong Li Thanks for the reply! Minor inconvenience, I'd rather the energy go in making the gaits more stable! when is that deep carpet walk mode coming :)

I'll try to debug a bit more during the weekend and maybe propose a fix.

Seventeen29

Aug 05, 2021

Replying to

@Knut Bear Hey! I've been working through the same issue where some serial commands are ignored, and I've also been trying to sift through the source code. Have you figured out what the "newCmdIdx" variable does?

Knut Bear

Jul 07, 2021

I wrote the following short program to try things out easily:

#!/usr/bin/python
import datetime
import sys
import tty
import termios

from ardSerial import wrapper

# read a single character on a unix system
# https://exceptionshub.com/python-read-a-single-character-from-the-user.html
def read_unix(count):
  ch = sys.stdin.read(count)
  return ch

fd = sys.stdin.fileno()
old_settings = termios.tcgetattr(fd)
tty.setraw(fd)

wrapper(['g',2])
wrapper(['kbalance',2])

while True:
  char = read_unix(1)
  
  if char == "q":
    break
  elif char == "a":
    wrapper(['kcrL',1])
  elif char == "s":
    wrapper(['bk',1])
  elif char == "d":
    wrapper(['kcrR',1])
  elif char == "w":
    wrapper(['kcrF',1])
 
termios.tcsetattr(fd, termios.TCSADRAIN, old_settings)
wrapper(['d',2])

I definitely get the log of the commands sent by ardSerial - and the balance/turn off commands work reliably, but the move skills do not - sometimes they work, sometimes the command gets ignored.

Position controller and its gains

Fixable!!

Issue connecting Bluetooth BittleX with Mac

Driving bittle from Raspberry PI