forked from tangger/lerobot
Extend reward classifier for multiple camera views (#626)
This commit is contained in:
@@ -161,6 +161,7 @@ from lerobot.common.robot_devices.control_utils import (
|
||||
log_control_info,
|
||||
record_episode,
|
||||
reset_environment,
|
||||
reset_follower_position,
|
||||
sanity_check_dataset_name,
|
||||
sanity_check_dataset_robot_compatibility,
|
||||
stop_recording,
|
||||
@@ -281,6 +282,9 @@ def record(
|
||||
robot.connect()
|
||||
listener, events = init_keyboard_listener(assign_rewards=assign_rewards)
|
||||
|
||||
if reset_follower:
|
||||
initial_position = robot.follower_arms["main"].read("Present_Position")
|
||||
|
||||
# Execute a few seconds without recording to:
|
||||
# 1. teleoperate the robot to move it in starting position if no policy provided,
|
||||
# 2. give times to the robot devices to connect and start synchronizing,
|
||||
|
||||
Reference in New Issue
Block a user