AI program for MBot Neo using OpenAI Whisper and Mycroft Adapt-Intent Parser.
NOTE: This project is under active development and it may not be suitable for your use. I make no gurantees that it will work on your robot or in your environment.
Currently the makeblock Python package is hard coded to look for a specific device identifier when doing it's auto connect.
In my local virtualenv I have modified the makeblock module as follows:
boards/base/init.py:
def connect(self,device=None,channel=None):
if type(device)==int:
channel = device
if channel is None:
channels = mlink.list()
if len(channels)>0:
self._dev = mlink.connect(channels[0])
return self.__setup(self._type,self._dev)
else:
self._device = mlink.connect(channel)
return self.__setup(self._type,self._dev)
if device is None:
print([a[2] for a in SerialPort.list()])
ports = [port[0] for port in SerialPort.list() if port[2] != 'n/a' and port[2].find('CNCA1')>0 ]
print(ports)
if len(ports)>0:
device = SerialPort(ports[0])
self._dev = device
return self.__setup(self._type,self._dev)The main change was putting CNCA1 instead of the port[2].find('1A86:7523')>0 that was there before.
CNCA1 is the identifier of my port from com0com.
In the future I will work to make this change included in the DBot code and not have to modify an open source pacage directly.
docker-compose up -d
Using the Python ble-serial package we can run a ble-scan:
Started general BLE scan
AC:0B:FB:21:A9:4A (RSSI=-62): Makeblock_LEAC0BFB21A94A
Finished general BLE scan
Then once we have the identifier (In this case AC:0B:FB:21:A9:4A) we can run a deep scan.
I use the following command to connect to the MBot Neo using ble-serial:
ble-serial -d AC:0B:FB:21:A9:4A -w 0000ffe3-0000-1000-8000-00805f9b34fb -r 0000ffe2-0000-1000-8000-00805f9b34fb -v