Source code for autopilot.tasks.children

Sub-tasks that serve as children to other tasks.

.. note::

    The Child agent will be formalized in an upcoming release, until then these classes
    remain relatively undocumented as their design will likely change.


from collections import OrderedDict as odict
from collections import deque

import autopilot.transform
from autopilot import prefs
from autopilot.hardware.gpio import Digital_Out
from autopilot.hardware.usb import Wheel
from autopilot.hardware import cameras
from autopilot.networking import Net_Node
from autopilot.core.loggers import init_logger
from autopilot.transform import transforms
from autopilot.hardware.i2c import I2C_9DOF
from autopilot.hardware.cameras import PiCamera
from autopilot.tasks import Task
from itertools import cycle
from queue import Empty, LifoQueue
import threading
import logging
from time import sleep

[docs]class Child(object): """Just a placeholder class for now to work with :func:`autopilot.get`"""
[docs]class Wheel_Child(Child): STAGE_NAMES = ['collect'] PARAMS = odict() PARAMS['fs'] = {'tag': 'Velocity Reporting Rate (Hz)', 'type': 'int'} PARAMS['thresh'] = {'tag': 'Distance Threshold', 'type': 'int'} HARDWARE = { "OUTPUT": Digital_Out, "WHEEL": Wheel } def __init__(self, stage_block=None, fs=10, thresh=100, **kwargs): super(Wheel_Child, self).__init__(**kwargs) self.fs = fs self.thresh = thresh self.hardware = {} self.hardware['OUTPUT'] = Digital_Out(prefs.get('HARDWARE')['OUTPUT']) self.hardware['WHEEL'] = Wheel(digi_out = self.hardware['OUTPUT'], fs = self.fs, thresh = self.thresh, mode = "steady") self.stages = cycle([self.noop]) self.stage_block = stage_block
[docs] def noop(self): # just fitting in with the task structure. self.stage_block.clear() return {}
[docs] def end(self): self.hardware['WHEEL'].release() self.stage_block.set()
[docs]class Video_Child(Child): PARAMS = odict() PARAMS['cams'] = {'tag': 'Dictionary of camera params, or list of dicts', 'type': ('dict', 'list')} def __init__(self, cams=None, stage_block = None, start_now=True, **kwargs): """ Args: cams (dict, list): Should be a dictionary of camera parameters or a list of dicts. Dicts should have, at least:: { 'type': 'string_of_camera_class', 'name': 'name_of_camera_in_task', 'param1': 'first_param' } """ super(Video_Child, self).__init__(**kwargs) if cams is None: Exception('Need to give us a cams dictionary!') self.cams = {} self.start_now = start_now if isinstance(cams, dict): try: cam_class = getattr(cameras, cams['type']) self.cams[cams['name']] = cam_class(**cams) # if start: # self.cams[cams['name']].capture() except AttributeError: AttributeError("Camera type {} not found!".format(cams['type'])) elif isinstance(cams, list): for cam in cams: try: cam_class = getattr(cameras, cam['type']) self.cams[cam['name']] = cam_class(**cam) # if start: # self.cams[cam['name']].capture() except AttributeError: AttributeError("Camera type {} not found!".format(cam['type'])) self.stages = cycle([self.noop]) self.stage_block = stage_block if self.start_now: self.start() # self.thread = threading.Thread(target=self._stream) # self.thread.daemon = True # self.thread.start()
[docs] def start(self): for cam in self.cams.values(): cam.capture()
[docs] def stop(self): for cam_name, cam in self.cams.items(): try: cam.release() except Exception as e: Warning('Couldnt release camera {},\n{}'.format(cam_name, e))
def _stream(self): self.node = Net_Node( "T_CHILD", upstream=prefs.get('NAME'), port=prefs.get('MSGPORT'), listens = {}, instance=True ) while True: for name, cam in self.cams.items(): try: frame, timestamp = cam.q.get_nowait() self.node.send(key='CONTINUOUS', value={, 'timestamp':timestamp}, repeat=False, flags={'MINPRINT':True}) except Empty: pass
[docs] def noop(self): # just fitting in with the task structure. self.stage_block.clear() return {}
# def start(self): # for cam in self.cams.values(): # cam.capture() # # def stop(self): # for cam in self.cams.values(): # cam.release()
[docs]class Transformer(Child): def __init__(self, transform, operation: str ="trigger", node_id = None, return_id = 'T', return_ip = None, return_port = None, return_key = None, router_port = None, stage_block = None, value_subset=None, forward_id=None, forward_ip=None, forward_port=None, forward_key=None, forward_what='both', **kwargs): """ Args: transform: operation (str): either * "trigger", where the last transform is a :class:`~autopilot.transform.transforms.Condition` and a trigger is returned to sender only when the return value of the transformation changes, or * "stream", where each result of the transformation is returned to sender return_id: return_ip: return_port: return_key: router_port (None, int): If not ``None`` (default), spawn the node with a route port to receieve stage_block: value_subset (str): Optional - subset a value from from a dict/list sent to :meth:`.l_process` forward_what (str): one of 'input', 'output', or 'both' (default) that determines what is forwarded **kwargs: """ super(Transformer, self).__init__(**kwargs) assert operation in ('trigger', 'stream', 'debug') self.operation = operation self._last_result = None if return_key is None: self.return_key = self.operation.upper() else: self.return_key = return_key self.return_id = return_id self.return_ip = return_ip self.return_port = return_port if self.return_port is None: self.return_port = prefs.get('MSGPORT') if node_id is None: self.node_id = f"{prefs.get('NAME')}_TRANSFORMER" else: self.node_id = node_id self.router_port = router_port self.forward_id = forward_id self.forward_ip = forward_ip self.forward_port = forward_port self.forward_key = forward_key self.forward_node = None self.forward_what = forward_what self.stage_block = stage_block self.stages = cycle([self.noop]) # self.input_q = LifoQueue() self.input_q = deque(maxlen=1) self.value_subset = value_subset self.logger = init_logger(self) self.process_thread = threading.Thread(target=self._process, args=(transform,)) self.process_thread.daemon = True self.process_thread.start()
[docs] def noop(self): # just fitting in with the task structure. self.stage_block.clear() return {}
def _process(self, transform): self.transform = autopilot.transform.make_transform(transform) self.node = Net_Node( self.node_id, upstream=self.return_id, upstream_ip=self.return_ip, port=self.return_port, router_port=self.router_port, listens = { 'CONTINUOUS': self.l_process }, instance=False ) if all([x is not None for x in (self.forward_id, self.forward_ip, self.forward_key, self.forward_port)]): self.forward_node = Net_Node( id=self.node_id, upstream=self.forward_id, upstream_ip=self.forward_ip, port=self.forward_port, listens={} ) self.node.send(self.return_id, 'STATE', value='READY') while True: try: # value = self.input_q.get_nowait() value = self.input_q.popleft() # except Empty: except IndexError: sleep(0.001) continue result = self.transform.process(value) self.node.logger.debug(f'Processed frame, result: {result}') if self.operation == "trigger": if result != self._last_result: self.node.send(self.return_id, self.return_key, result) if self.forward_node is not None: self.forward(value, result) self._last_result = result elif self.operation == 'stream': # FIXME: Another key that's not TRIGGER self.node.send(self.return_id, self.return_key, result) if self.forward_node is not None: self.forward(value, result) elif self.operation == 'debug': pass
[docs] def l_process(self, value): # get array out of value # FIXME hack for dlc self.node.logger.debug('Received and queued processing!') # self.input_q.put_nowait(value['MAIN']) if self.value_subset: value = value[self.value_subset] self.input_q.append(value)
[docs] def forward(self, input=None, output=None): if self.forward_what == 'both': self.forward_node.send(self.forward_id, self.forward_key, {'input':input,'output':output},flags={'MINPRINT':True,'NOREPEAT':True}) elif self.forward_what == 'input': self.forward_node.send(self.forward_id, self.forward_key, input,flags={'MINPRINT':True,'NOREPEAT':True}) elif self.forward_what == 'output': self.forward_node.send(self.forward_id, self.forward_key, output,flags={'MINPRINT':True,'NOREPEAT':True}) else: raise ValueError("forward_what must be one of 'input', 'output', or 'both'")