Add async_read and async_write (it doesnt work)

This commit is contained in:
Remi Cadene
2024-07-05 18:09:29 +02:00
parent 6e77a399a2
commit 3ff789c181
6 changed files with 243 additions and 62 deletions

View File

@@ -1,4 +1,5 @@
import argparse
import math
import time
from dataclasses import dataclass, replace
from pathlib import Path
@@ -8,6 +9,7 @@ import cv2
import numpy as np
from lerobot.common.robot_devices.cameras.utils import save_color_image
from lerobot.common.utils.utils import capture_timestamp_utc
def find_camera_indices(raise_when_empty=False, max_index_search_range=60):
@@ -119,9 +121,9 @@ class OpenCVCamera:
self.camera = None
self.is_connected = False
self.t = Thread(target=self.capture_image_loop, args=())
self.t.daemon = True
self._color_image = None
self.threads = {}
self.results = {}
self.logs = {}
def connect(self):
if self.is_connected:
@@ -161,7 +163,7 @@ class OpenCVCamera:
actual_width = self.camera.get(cv2.CAP_PROP_FRAME_WIDTH)
actual_height = self.camera.get(cv2.CAP_PROP_FRAME_HEIGHT)
if self.fps and self.fps != actual_fps:
if self.fps and not math.isclose(self.fps, actual_fps, rel_tol=1e-3):
raise OSError(
f"Can't set {self.fps=} for camera {self.camera_index}. Actual value is {actual_fps}."
)
@@ -175,12 +177,13 @@ class OpenCVCamera:
)
self.is_connected = True
self.t.start()
def capture_image(self, temporary_color: str | None = None) -> np.ndarray:
def read(self, temporary_color: str | None = None) -> np.ndarray:
if not self.is_connected:
self.connect()
start_time = time.perf_counter()
ret, color_image = self.camera.read()
if not ret:
raise OSError(f"Can't capture color image from camera {self.camera_index}.")
@@ -196,20 +199,41 @@ class OpenCVCamera:
if requested_color == "rgb":
color_image = cv2.cvtColor(color_image, cv2.COLOR_BGR2RGB)
# log the number of seconds it took to read the image
self.logs["delta_timestamp_s"] = time.perf_counter() - start_time
# log the utc time at which the image was received
self.logs["timestamp_utc"] = capture_timestamp_utc()
return color_image
def capture_image_loop(self):
def read_loop(self):
while True:
self._color_image = self.capture_image()
self.results["color_image"] = self.read()
def read(self):
while self._color_image is None:
time.sleep(0.1)
return self._color_image
def async_read(self):
if "read" not in self.threads:
self.threads["read"] = Thread(target=self.read_loop, args=())
self.threads["read"].daemon = True
self.threads["read"].start()
num_tries = 0
while "color_image" not in self.results:
num_tries += 1
time.sleep(1/self.fps)
if num_tries > self.fps:
if self.threads["read"].ident is None and not self.threads["read"].is_alive():
raise Exception("The thread responsible for `self.async_read()` took too much time to start. There might be an issue. Verify that `self.threads[\"read\"].start()` has been called.")
return self.results["color_image"] #, self.logs["timestamp_utc"]
def disconnect(self):
if getattr(self, "camera", None):
self.camera.release()
for name in self.threads:
if self.threads[name].is_alive():
# wait for the thread to finish
self.threads[name].join()
def __del__(self):
self.disconnect()
@@ -217,8 +241,10 @@ class OpenCVCamera:
def save_images_config(config: OpenCVCameraConfig, out_dir: Path):
cameras = []
print(f"Available camera indices: {OpenCVCamera.AVAILABLE_CAMERAS_INDICES}")
for camera_idx in OpenCVCamera.AVAILABLE_CAMERAS_INDICES:
available_cam_ids = find_camera_indices()
print(f"Available camera indices: {available_cam_ids}")
for camera_idx in available_cam_ids:
camera = OpenCVCamera(camera_idx, config)
cameras.append(camera)

View File

@@ -1,5 +1,8 @@
import enum
from copy import deepcopy
from queue import Queue
from threading import Thread
import time
import numpy as np
from dynamixel_sdk import (
@@ -14,6 +17,8 @@ from dynamixel_sdk import (
PortHandler,
)
from lerobot.common.utils.utils import capture_timestamp_utc
PROTOCOL_VERSION = 2.0
BAUD_RATE = 1_000_000
TIMEOUT_MS = 1000
@@ -149,6 +154,15 @@ def get_group_sync_key(data_name, motor_names):
group_key = f"{data_name}_" + "_".join(motor_names)
return group_key
def get_thread_name(fn_name, data_name, motor_names):
group_key = get_group_sync_key(data_name, motor_names)
thread_name = f"{fn_name}_{group_key}"
return thread_name
def get_log_name(var_name, fn_name, data_name, motor_names):
thread_name = get_thread_name(fn_name, data_name, motor_names)
log_name = f"{var_name}_{thread_name}"
return log_name
class TorqueMode(enum.Enum):
ENABLED = 1
@@ -197,6 +211,11 @@ class DynamixelMotorsBus:
self.calibration = None
self.threads = {}
self.queues = {}
self.results = {}
self.logs = {}
@property
def motor_names(self) -> list[int]:
return list(self.motors.keys())
@@ -239,6 +258,8 @@ class DynamixelMotorsBus:
return values
def read(self, data_name, motor_names: list[str] | None = None):
start_time = time.perf_counter()
if motor_names is None:
motor_names = self.motor_names
@@ -259,7 +280,12 @@ class DynamixelMotorsBus:
for idx in motor_ids:
self.group_readers[group_key].addParam(idx)
comm = self.group_readers[group_key].txRxPacket()
NUM_TRIES = 10
for _ in range(NUM_TRIES):
comm = self.group_readers[group_key].txRxPacket()
if comm == COMM_SUCCESS:
break
if comm != COMM_SUCCESS:
raise ConnectionError(
f"Read failed due to communication error on port {self.port} for group_key {group_key}: "
@@ -283,15 +309,30 @@ class DynamixelMotorsBus:
if data_name in CONVERT_POSITION_TO_ANGLE_REQUIRED:
values = motor_position_to_angle(values)
# log the number of seconds it took to read the data from the motors
delta_ts_name = get_log_name("delta_timestamp_s", "read", data_name, motor_names)
self.logs[delta_ts_name] = time.perf_counter() - start_time
# log the utc time at which the data was received
ts_utc_name = get_log_name("timestamp_utc", "read", data_name, motor_names)
self.logs[ts_utc_name] = capture_timestamp_utc()
return values
def write(self, data_name, values: int | float | np.ndarray, motor_names: str | list[str] | None = None):
start_time = time.perf_counter()
if motor_names is None:
motor_names = self.motor_names
if isinstance(motor_names, str):
motor_names = [motor_names]
if isinstance(values, (int, float, np.integer)):
values = [int(values)] * len(motor_names)
values = np.array(values)
motor_ids = []
models = []
for name in motor_names:
@@ -299,11 +340,6 @@ class DynamixelMotorsBus:
motor_ids.append(motor_idx)
models.append(model)
if isinstance(values, (int, float, np.integer)):
values = [int(values)] * len(motor_ids)
values = np.array(values)
if data_name in CONVERT_POSITION_TO_ANGLE_REQUIRED:
values = motor_angle_to_position(values)
@@ -361,6 +397,99 @@ class DynamixelMotorsBus:
f"{self.packet_handler.getTxRxResult(comm)}"
)
# log the number of seconds it took to write the data to the motors
delta_ts_name = get_log_name("delta_timestamp_s", "write", data_name, motor_names)
self.logs[delta_ts_name] = time.perf_counter() - start_time
# TODO(rcadene): should we log the time before sending the write command?
# log the utc time when the write has been completed
ts_utc_name = get_log_name("timestamp_utc", "write", data_name, motor_names)
self.logs[ts_utc_name] = capture_timestamp_utc()
def read_loop(self, data_name, motor_names: list[str] | None = None):
while True:
thread_name = get_thread_name("read", data_name, motor_names)
self.results[thread_name] = self.read(data_name, motor_names)
def async_read(self, data_name, motor_names: list[str] | None = None):
if motor_names is None:
motor_names = self.motor_names
if isinstance(motor_names, str):
motor_names = [motor_names]
thread_name = get_thread_name("read", data_name, motor_names)
if thread_name not in self.threads:
self.threads[thread_name] = Thread(target=self.read_loop, args=(data_name, motor_names))
self.threads[thread_name].daemon = True
self.threads[thread_name].start()
FPS = 200
num_tries = 0
while thread_name not in self.results:
num_tries += 1
time.sleep(1 / FPS)
if num_tries > FPS:
if self.threads[thread_name].ident is None and not self.threads[thread_name].is_alive():
raise Exception(f"The thread responsible for `self.async_read({data_name}, {motor_names})` took too much time to start. There might be an issue. Verify that `self.threads[thread_name].start()` has been called.")
# ts_utc_name = get_log_name("timestamp_utc", "read", data_name, motor_names)
return self.results[thread_name] #, self.logs[ts_utc_name]
def write_loop(self, data_name, queue: Queue, motor_names: list[str] | None = None):
while True:
values = queue.get()
if values is None: # A way to terminate the thread
break
self.write(data_name, values, motor_names)
queue.task_done()
def async_write(self, data_name, values: int | float | np.ndarray, motor_names: str | list[str] | None = None):
if motor_names is None:
motor_names = self.motor_names
if isinstance(motor_names, str):
motor_names = [motor_names]
if isinstance(values, (int, float, np.integer)):
values = [int(values)] * len(motor_names)
values = np.array(values)
thread_name = get_thread_name("write", data_name, motor_names)
ts_utc_name = get_log_name("timestamp_utc", "write", data_name, motor_names)
if thread_name not in self.threads:
self.queues[thread_name] = Queue()
self.threads[thread_name] = Thread(target=self.write_loop, args=(data_name, self.queues[thread_name], motor_names))
self.threads[thread_name].daemon = True
self.threads[thread_name].start()
self.queues[thread_name].put(values)
FPS = 200
num_tries = 0
while ts_utc_name not in self.logs:
num_tries += 1
time.sleep(1 / FPS)
if num_tries > FPS:
if self.threads[thread_name].ident is None and not self.threads[thread_name].is_alive():
raise Exception(f"The thread responsible for `self.async_write({data_name}, {values}, {motor_names})` took too much time to start. There might be an issue. Verify that `self.threads[thread_name].start()` has been called.")
return self.logs[ts_utc_name]
def __del__(self):
for thread_name in self.queues:
# Send value that corresponds to `break` logic
self.queues[thread_name].put(None)
self.queues[thread_name].join()
for thread_name in self.queues:
self.threads[thread_name].join()
# TODO(rcadene): find a simple way to exit threads created by async_read
# def read(self, data_name, motor_name: str):
# motor_idx, model = self.motors[motor_name]
# addr, bytes = self.model_ctrl_table[model][data_name]

View File

@@ -34,7 +34,8 @@ def make_robot(name):
),
},
cameras={
"main": OpenCVCamera(1, fps=30, width=640, height=480),
"macbookpro": OpenCVCamera(1, fps=30, width=640, height=480),
"iphone": OpenCVCamera(2, fps=30, width=640, height=480),
},
)
else:

View File

@@ -244,6 +244,9 @@ class KochRobot:
self.follower_arms = self.config.follower_arms
self.cameras = self.config.cameras
self.async_read = False
self.async_write = False
def init_teleop(self):
if self.calibration_path.exists():
# Reset all arms before setting calibration
@@ -265,6 +268,7 @@ class KochRobot:
for name in self.follower_arms:
self.follower_arms[name].set_calibration(calibration[f"follower_{name}"])
self.follower_arms[name].write("Torque_Enable", 1)
self.follower_arms[name].write("Torque_Enable", 1)
for name in self.leader_arms:
self.leader_arms[name].set_calibration(calibration[f"leader_{name}"])
@@ -300,7 +304,10 @@ class KochRobot:
# Prepare to assign the positions of the leader to the follower
leader_pos = {}
for name in self.leader_arms:
leader_pos[name] = self.leader_arms[name].read("Present_Position")
if self.async_read:
leader_pos[name] = self.leader_arms[name].async_read("Present_Position")
else:
leader_pos[name] = self.leader_arms[name].read("Present_Position")
follower_goal_pos = {}
for name in self.leader_arms:
@@ -308,7 +315,10 @@ class KochRobot:
# Send action
for name in self.follower_arms:
self.follower_arms[name].write("Goal_Position", follower_goal_pos[name])
if self.async_write:
self.follower_arms[name].async_write("Goal_Position", follower_goal_pos[name])
else:
self.follower_arms[name].write("Goal_Position", follower_goal_pos[name])
# Early exit when recording data is not requested
if not record_data:
@@ -317,7 +327,10 @@ class KochRobot:
# Read follower position
follower_pos = {}
for name in self.follower_arms:
follower_pos[name] = self.follower_arms[name].read("Present_Position")
if self.async_read:
follower_pos[name] = self.follower_arms[name].async_read("Present_Position")
else:
follower_pos[name] = self.follower_arms[name].read("Present_Position")
# Create state by concatenating follower current position
state = []
@@ -336,7 +349,7 @@ class KochRobot:
# Capture images from cameras
images = {}
for name in self.cameras:
images[name] = self.cameras[name].read()
images[name] = self.cameras[name].async_read()
# Populate output dictionnaries and format to pytorch
obs_dict, action_dict = {}, {}
@@ -351,7 +364,10 @@ class KochRobot:
# Read follower position
follower_pos = {}
for name in self.follower_arms:
follower_pos[name] = self.follower_arms[name].read("Present_Position")
if self.async_read:
follower_pos[name] = self.follower_arms[name].async_read("Present_Position")
else:
follower_pos[name] = self.follower_arms[name].read("Present_Position")
# Create state by concatenating follower current position
state = []
@@ -363,7 +379,7 @@ class KochRobot:
# Capture images from cameras
images = {}
for name in self.cameras:
images[name] = self.cameras[name].read()
images[name] = self.cameras[name].async_read()
# Populate output dictionnaries and format to pytorch
obs_dict = {}
@@ -383,4 +399,7 @@ class KochRobot:
from_idx = to_idx
for name in self.follower_arms:
self.follower_arms[name].write("Goal_Position", follower_goal_pos[name].astype(np.int32))
if self.async_write:
self.follower_arms[name].async_write("Goal_Position", follower_goal_pos[name].astype(np.int32))
else:
self.follower_arms[name].write("Goal_Position", follower_goal_pos[name].astype(np.int32))

View File

@@ -17,7 +17,7 @@ import logging
import os.path as osp
import random
from contextlib import contextmanager
from datetime import datetime
from datetime import datetime, timezone
from pathlib import Path
from typing import Any, Generator
@@ -172,3 +172,7 @@ def print_cuda_memory_usage():
print("Maximum GPU Memory Allocated: {:.2f} MB".format(torch.cuda.max_memory_allocated(0) / 1024**2))
print("Current GPU Memory Reserved: {:.2f} MB".format(torch.cuda.memory_reserved(0) / 1024**2))
print("Maximum GPU Memory Reserved: {:.2f} MB".format(torch.cuda.max_memory_reserved(0) / 1024**2))
def capture_timestamp_utc():
return datetime.now(timezone.utc)

View File

@@ -101,7 +101,7 @@ def save_image(img_tensor, key, frame_index, episode_index, videos_dir):
def busy_wait(seconds):
# Significantly more accurate than `time.sleep`, and mendatory for our use case,
# but it consumes CPU cycles.
# TODO(rcadene): find an alternative
# TODO(rcadene): find an alternative: from python 11, time.sleep is precise
end_time = time.perf_counter() + seconds
while time.perf_counter() < end_time:
pass
@@ -156,42 +156,46 @@ def record_dataset(
videos_dir = local_dir / "videos"
videos_dir.mkdir(parents=True, exist_ok=True)
start_time = time.perf_counter()
is_warmup_print = False
is_record_print = False
ep_dicts = []
# Save images using threads to reach high fps (30 and more)
# Using `with` ensures the program exists smoothly if an execption is raised.
with concurrent.futures.ThreadPoolExecutor() as executor:
timestamp = 0
start_time = time.perf_counter()
is_warmup_print = False
while timestamp < warmup_time_s:
if not is_warmup_print:
print("Warming up by skipping frames")
os.system('say "Warmup" &')
is_warmup_print = True
now = time.perf_counter()
observation, action = robot.teleop_step(record_data=True)
dt_s = time.perf_counter() - now
busy_wait(1 / fps - dt_s)
dt_s = time.perf_counter() - now
print(f"Latency (ms): {dt_s * 1000:.2f}\tFrequency: {1 / dt_s:.2f} (Warmup)")
timestamp = time.perf_counter() - start_time
ep_dicts = []
for episode_index in range(num_episodes):
ep_dict = {}
frame_index = 0
while True:
if not is_warmup_print:
print("Warming up by skipping frames")
os.system('say "Warmup"')
is_warmup_print = True
now = time.perf_counter()
observation, action = robot.teleop_step(record_data=True)
timestamp = time.perf_counter() - start_time
if timestamp < warmup_time_s:
dt_s = time.perf_counter() - now
busy_wait(1 / fps - dt_s)
dt_s = time.perf_counter() - now
print(f"Latency (ms): {dt_s * 1000:.2f}\tFrequency: {1 / dt_s:.2f} (Warmup)")
continue
timestamp = 0
start_time = time.perf_counter()
is_record_print = False
while timestamp < episode_time_s:
if not is_record_print:
print("Recording")
os.system(f'say "Recording episode {episode_index}"')
print(f"Recording episode {episode_index}")
os.system(f'say "Recording episode {episode_index}" &')
is_record_print = True
now = time.perf_counter()
observation, action = robot.teleop_step(record_data=True)
image_keys = [key for key in observation if "image" in key]
not_image_keys = [key for key in observation if "image" not in key]
@@ -216,11 +220,9 @@ def record_dataset(
dt_s = time.perf_counter() - now
print(f"Latency (ms): {dt_s * 1000:.2f}\tFrequency: {1 / dt_s:.2f}")
if timestamp > episode_time_s - warmup_time_s:
break
timestamp = time.perf_counter() - start_time
print("Encoding to `LeRobotDataset` format")
os.system('say "Encoding"')
print("Encoding images to videos")
num_frames = frame_index
@@ -271,7 +273,7 @@ def record_dataset(
for key in image_keys:
time.sleep(10)
tmp_imgs_dir = videos_dir / f"{key}_episode_{episode_index:06d}"
# shutil.rmtree(tmp_imgs_dir)
shutil.rmtree(tmp_imgs_dir)
lerobot_dataset = LeRobotDataset.from_preloaded(
repo_id=repo_id,