Today I got the kids involved making - and we had some fun getting the motor cables the right way. However, the stall outs, or lock ups troubled me.
Motor lock out problem
The piconzero on this (and tankbot) kept stopping - and it could even stop with the motors running. This was disturbing.
I was researching stuff on motor stall currents, and brushing up my physics - what had I missed? The robot could fail only seconds in, or a while in.
After checking on social media, and then reading the piconzero docs - it turns out that older pi raspbian images had i2c stability issues. So I started a dist-upgrade on it. This seemed like a great plan…
Bad news then - I knocked out the power cable (external wall power) while doing the update!
The image I had was not going to work again. I’d luckily backed up my own code before doing so - so nothing lost but time.
And then “approx_eng”, a creator of Raspberry Pi and robotics libraries, often present at piwars, came to the rescue with an image he had with BlueZ and sixaxis control on board, along with openCV. Which will save my compiling again.
My old image used Python3 - so after using his image, there was some adapting to do.
First - smbus needed to be smbus2 - but this was because I had overlooked I was in a virtual env - once I’d adapted that, the motor test worked. However, @biglesp had python3 adaptations of the piconzero code I could use now.
I got it to drive again! It was no longer hanging, but I was getting squealing motor stalls.
Motor stalls with squealing
My next thought was to try taming it from full spin to just braking a tread at a time.
So I make a tankTest.py module adapted from motorTest.py - using the numeric keypad like tank drive controls:
| Motor | Left | Right |
|---|---|---|
| Forward | 7 | 9 |
| Stop | 4 | 6 |
| Reverse | 1 | 3 |
This was then driving brilliantly - few stalls, no piconzero cutouts. Stability seems good.
Controller
This Raspberry PI SD card image from @approx_eng comes with BlueZ (bluetooth linux tool) preinstalled - so lets try getting the controller live.
- First - get the pad to pair/trusted in BlueControl.
- Then make a bit of code to read the pad - plan - 2 Sticks for left track/right track. Try to get button reads - start, select, 4 main buttons. Triggers later.
- This should just print the read values.
- Adapt the tankTest to this.
- Test drive - and video.
- Publish this on orionrobots blog (filtering out real uuid’s)
Pairing
Powered up the pi - powered from PC via USB to preserve robot battery power. Started up bluetoothctl as pi user.
Lets plug in the pad to see what happens - according to @approx_eng, this kernel and bluetoothctl has a plugin for registering the sixaxis.
I plug in and see this:
[NEW] Device 00:19:C1:DE:AD:BE Sony PLAYSTATION(R)3 Controller
[DEL] Device 00:19:C1:DE:AD:BE Sony PLAYSTATION(R)3 ControllerWhy is it NEW then DEL?
According to Approx_eng I should expect this:
The guide doesn’t assume you’ve rebuilt BlueZ from source! As I did, you should be able to run bluetoothctl, plug the controller in, say ‘yes’ to the prompt, then unplug and press the PS button.
Hmm - no prompt. Lets try unplug, then press the PS button. Flashing lights on the PAD, nothing in bluetoothctl.
Do Need to be root on the Pi? Same - NEW/DEL combo. Nothing further from BlueToothCtl.
Time to dive into DMesg (where linux tends to keep hardware log events).
[ 232.228847] usb 1-1.4: USB disconnect, device number 4
[ 268.939861] usb 1-1.4: new full-speed USB device number 5 using dwc_otg
[ 269.097846] usb 1-1.4: New USB device found, idVendor=054c, idProduct=0268
[ 269.097859] usb 1-1.4: New USB device strings: Mfr=1, Product=2, SerialNumber =0
[ 269.097869] usb 1-1.4: Product: PLAYSTATION(R)3 Controller
[ 269.097878] usb 1-1.4: Manufacturer: Sony
[ 269.182343] input: Sony PLAYSTATION(R)3 Controller Motion Sensors as /devices /platform/soc/3f980000.usb/usb1/1-1/1-1.4/1-1.4:1.0/0003:DEAD:BEEF.0002/input/in put3
[ 269.250204] input: Sony PLAYSTATION(R)3 Controller as /devices/platform/soc/3 f980000.usb/usb1/1-1/1-1.4/1-1.4:1.0/0003:DEAD:BEEF.0002/input/input2
[ 269.250997] sony 0003:DEAD:BEEF.0002: input,hiddev96,hidraw0: USB HID v81.11 Joystick [Sony PLAYSTATION(R)3 Controller] on usb-3f980000.usb-1.4/input0
[ 270.380141] Under-voltage detected! (0x00050005)
[ 276.621037] Voltage normalised (0x00000000)
[ 284.202314] usb 1-1.4: USB disconnect, device number 5Hmm - We have the device - when plugged in. But where is the bluetooth stuff? Need to probe more into how this image is working. The under voltage has me worried. But first - is there a bluetooth log? Lets check /var/log as root - and see what I have:
-rw-r--r-- 1 root root 7675 Mar 31 23:28 alternatives.log
drwxr-xr-x 2 root root 4096 Mar 31 23:28 apt
-rw-r----- 1 root adm 121369 Apr 1 14:03 auth.log
-rw-r--r-- 1 root root 4925 Apr 1 11:30 boot.log
-rw-r--r-- 1 root root 0 Mar 13 22:53 bootstrap.log
-rw------- 1 root utmp 768 Mar 13 23:11 btmp
-rw-r----- 1 root adm 472905 Apr 1 13:53 daemon.log
-rw-r----- 1 root adm 26173 Apr 1 11:29 debug
-rw-r--r-- 1 root root 263885 Mar 31 23:28 dpkg.log
-rw-r--r-- 1 root root 0 Mar 13 22:53 faillog
-rw-r--r-- 1 root root 751 Mar 19 18:08 fontconfig.log
-rw-r----- 1 root adm 710830 Apr 1 13:53 kern.log
-rw-rw-r-- 1 root utmp 292292 Apr 1 11:30 lastlog
-rw-r----- 1 root adm 679805 Apr 1 13:53 messages
drwxr-x--- 2 root adm 4096 Nov 20 22:24 samba
-rw-r----- 1 root adm 1216548 Apr 1 13:53 syslog
-rw-r----- 1 root adm 22170 Apr 1 13:53 user.log
-rw-rw-r-- 1 root utmp 85632 Apr 1 11:30 wtmpOkay - start with messages and syslog. I’m using the less command and shift-g to get to the end of a log.
Apr 1 13:53:44 skittlebot kernel: [ 268.939861] usb 1-1.4: new full-speed USB device number 5 using dwc_otg
Apr 1 13:53:44 skittlebot kernel: [ 269.097846] usb 1-1.4: New USB device found, idVendor=054c, idProduct=0268
Apr 1 13:53:44 skittlebot kernel: [ 269.097859] usb 1-1.4: New USB device strings: Mfr=1, Product=2, SerialNumber=0
Apr 1 13:53:44 skittlebot kernel: [ 269.097869] usb 1-1.4: Product: PLAYSTATION(R)3 Controller
Apr 1 13:53:44 skittlebot kernel: [ 269.097878] usb 1-1.4: Manufacturer: Sony
Apr 1 13:53:44 skittlebot kernel: [ 269.182343] input: Sony PLAYSTATION(R)3 Controller Motion Sensors as /devices/platform/soc/3f980000.usb/usb1/1-1/1-1.4/1-1.4:1.0/0003:DEAD:BEEF.0002/input/input3
Apr 1 13:53:44 skittlebot kernel: [ 269.250204] input: Sony PLAYSTATION(R)3 Controller as /devices/platform/soc/3f980000.usb/usb1/1-1/1-1.4/1-1.4:1.0/0003:DEAD:BEEF.0002/input/input2
Apr 1 13:53:44 skittlebot kernel: [ 269.250997] sony 0003:DEAD:BEEF.0002: input,hiddev96,hidraw0: USB HID v81.11 Joystick [Sony PLAYSTATION(R)3 Controller] on usb-3f980000.usb-1.4/input0
Apr 1 13:53:44 skittlebot mtp-probe: checking bus 1, device 5: "/sys/devices/platform/soc/3f980000.usb/usb1/1-1/1-1.4"
Apr 1 13:53:44 skittlebot mtp-probe: bus: 1, device: 5 was not an MTP device
Apr 1 13:53:52 skittlebot kernel: [ 276.621037] Voltage normalised (0x00000000)What is “mtp-probe”? Media-transfer-protocol - okay, that is irrelevant. Try syslog.
Apr 1 13:53:44 skittlebot kernel: [ 269.182343] input: Sony PLAYSTATION(R)3 Controller Motion Sensors as /devices/platform/soc/3f980000.usb/usb1/1-1/1-1.4/1-1.4:1.0/0003:DEAD:BEEF.0002/input/input3
Apr 1 13:53:44 skittlebot kernel: [ 269.250204] input: Sony PLAYSTATION(R)3 Controller as /devices/platform/soc/3f980000.usb/usb1/1-1/1-1.4/1-1.4:1.0/0003:DEAD:BEEF.0002/input/input2
Apr 1 13:53:44 skittlebot kernel: [ 269.250997] sony 0003:DEAD:BEEF.0002: input,hiddev96,hidraw0: USB HID v81.11 Joystick [Sony PLAYSTATION(R)3 Controller] on usb-3f980000.usb-1.4/input0
Apr 1 13:53:44 skittlebot mtp-probe: checking bus 1, device 5: "/sys/devices/platform/soc/3f980000.usb/usb1/1-1/1-1.4"
Apr 1 13:53:44 skittlebot mtp-probe: bus: 1, device: 5 was not an MTP device
Apr 1 13:53:44 skittlebot bluetoothd[433]: sixaxis: compatible device connected: Sony PLAYSTATION(R)3 Controller (054C:0268 /sys/devices/platform/soc/3f980000.usb/usb1/1-1/1-1.4/1-1.4:1.0/0003:DEAD:BEEF.0002/hidraw/hidraw0)
Apr 1 13:53:44 skittlebot bluetoothd[433]: sixaxis: setting up new device
Apr 1 13:53:44 skittlebot bluetoothd[433]: Authentication attempt without agentAh-hah “Authentication attempt without agent”. Sixaxis is recognising the device. So what is this “authentication attempt without agent” issue? Lets bring up bluetoothctl again.
bluetoothctl
Agent registered
[bluetooth]# agent on
Agent is already registeredHmm. Lets google for that message. Well the hits suggest I need to trust the device. Lets try that - it didn’t work last night. Plug in, then do the trust.
[NEW] Device 00:19:BA:DF:EE:D1 Sony PLAYSTATION(R)3 Controller
[DEL] Device 00:19:BA:DF:EE:D1 Sony PLAYSTATION(R)3 Controller
[bluetooth]# trust 00:19:BA:DF:EE:D1
Device 00:19:BA:DF:EE:D1 not availableWell why is it then not available? Lets try some more with bluetoothctl - maybe it’s not properly discoverable by the sixaxis.
[bluetooth]# default-agent
Default agent request successful
[bluetooth]# discoverable on
Changing discoverable on succeeded
[CHG] Controller BE:DB:EB:6B:ED:ED Discoverable: yes
[bluetooth]# pairable on
Changing pairable on succeeded
[bluetooth]# scan onAnd now the samsung TV - even in standby is flooding my console. How annoying. Lets unplug and reattach the sixaxis.
[NEW] Device 00:19:BA:DF:EE:D1 Sony PLAYSTATION(R)3 Controller
Authorize service
[blue1m[agent] Authorize service 00001124-0000-1000-8000-0080deadbeef (yes/no):yes
[CHG] Device 00:19:BA:DF:EE:D1 Trusted: yes
[CHG] Device 00:19:BA:DF:EE:D1 UUIDs: 00001124-0000-1000-8000-0080deadbeefThat’s it! Now I may need default scripts to make this happen on startup… Now unplug - press PS - and it is controller 1! Success.
root@skittlebot:/var/log# ls /dev/input
event0 event1 js0 miceAccording to The DualShock PS 3 Input Library Guide:
If a new device has appeared here then congratulations, you have successfully paired your dongle and SixAxis controller. This will persist across reboots, so from now on you can just connect by pressing the PS button on the controller. Pressing and holding this button will shut the controller down - at the moment there’s no timeout so be sure to turn the controller off when you’re not going to be using it for a while.
Now lets get some python code reading it. Starting here: https://approxeng.github.io/approxeng.input/simpleusage.html. I have some old code in my git repo - for doing this in pygame. But no - I think I would like to try it with this stand alone lib. I’ve heard good things about the input library. Lets start off with his sample code - understanding the parts of it.
Nice pertinent bit of information here:
The connected property on the joystick object indicates whether the underlying device is connected or not - if, for example, you have a controller that goes out of range, runs out of batteries, or is turned off while in use this will be set to False and you can handle the case correctly (if using this code in a robot, this would be an excellent time to turn of all your motors, for instance!)
Ok some code added to read the two sticks Y position and print it.
print("Left track", joystick.ly, "Right track", joystick.ry)Lets run this… Ah - no module approxeng - can I pip install it?
(env3) pi@skittlebot:~/piwars-bot $ pip install approxeng.input
Collecting approxeng.input
Downloading https://www.piwheels.org/simple/approxeng-input/approxeng.input-2.0.4-py3-none-any.whl
...Oh yes…
Now output - trying the left and right sticks - but not getting what I’d expect here:
Left track 0 Right track 0
[W 180401 14:32:35 __init__:295] Unknown axis code 4, value 127
[W 180401 14:32:35 __init__:295] Unknown axis code 4, value 126What does this mean? Wait - 2.0.4?
Now using the controller example from https://approxeng.github.io/approxeng.input/examples/show_controls.html - and the axis are weird. Time to talk with other robot builders.
Trying to remove “pi” from the “input” group. Then log out, back in and try again. No - nothing. Try a reboot. Without the pad - input shows only “mice”. With it - I see 4 items:
event0 event1 js0 miceShow controls does nothing - sits waiting for controller. ls -l shows the story:
(env3) pi@skittlebot:~/piwars-bot $ ls -l /dev/input/
total 0
crw-rw---- 1 root input 13, 64 Apr 1 15:23 event0
crw-rw---- 1 root input 13, 65 Apr 1 15:23 event1
crw-rw---- 1 root input 13, 0 Apr 1 15:23 js0
crw-rw---- 1 root input 13, 63 Apr 1 15:22 miceI spoke to the approx_eng guy. With Helena’s help - and used evtest - which shows event sources and allows you to look at them.
This showed nothing when not in the input group, and 2 event sources when in the group.
(env3) pi@skittlebot:~ $ evtest
No device specified, trying to scan all of /dev/input/event*
Not running as root, no devices may be available.
Available devices:
/dev/input/event0: Sony PLAYSTATION(R)3 Controller Motion Sensors
/dev/input/event1: Sony PLAYSTATION(R)3 Controllerevent0- is the motion sensing - but no buttons.event1- is the stick with it’s buttons.
So my next thought was to look at how the bindings work.
How do bindings work?
To do this - I dug in here: https://approxeng.github.io/approxeng.input/api/selectbinder.html And we wrote some code:
from approxeng.input.selectbinder import bind_controller
from approxeng.input.dualshock3 import DualShock3
from approxeng.input.controllers import find_single_controller
devices, controller, p = find_single_controller(controller_class = DualShock3)
import pdb
pdb.set_trace()This way we could examine the output of single controller - with a view to changing the event source.
The set_trace was to see it.
This was fruitful. When it ran - I got this:
(Pdb) devices
[InputDevice('/dev/input/event1'), InputDevice('/dev/input/event0')]
(Pdb) controller
Sony DualShock3 (Playstation 3) controllerSo - can we remove event 0? Event 1 is first - which is the right device. Lets try this to remove event0 from the list:
devices = [devices[0]]Lets try binding and printing;
with ControllerResource(devices = devices, controller = controller) as joystick:
while joystick.connected:
# Do stuff with your joystick here!
print("Left track {0:.2} Right track {1:.2}".format(joystick.ly*1.0, joystick.ry*1.0))
# Joystick disconnected...
print('Connection to joystick lost')This kinda works. We are getting correct values for the first stick.
Seeing the left axis only
The second stick - r - is bound still to rT! So - lets try the show thing again.
Well - I can see the l axis only - no buttons. Lets try my own control class… Tomorrow - lets pick up https://github.com/ApproxEng/approxeng.input/blob/master/src/python/approxeng/input/dualshock3.py and see if I can adapt it. Or see if I get a code drop from approx-eng.
Alternative plans
Plan B Buy one of these - https://thepihut.com/collections/raspberry-pi-gaming/products/raspberry-pi-compatible-wireless-gamepad-controller (or rock candy - but the dongle is soooo big)
Plan C the microbit radio hack.
Lets have a new day…
Next stages for the PiWars robot
- Fitting camera
- Fitting distance sensors
- Getting my SkittleBot code adapted for both the driving style and new opencv.
- Making the distance sensor code.
- Tidying up - cables a mess - my plan is to use PC drive power adaptors.
- Making a cover - not sure how much time, but this is Helena’s most wanted task.