fixed errors in rktl_autonomy

src/rktl_autonomy/nodes/plotter

@@ -41,7 +41,7 @@ class Plotter(rclpy.Node):
         #rospy.Subscriber('~log', DiagnosticStatus, self.progress_cb)
         self.create_subscription(DiagnosticStatus, '~log', self.progress_cb, qos_profile=10)
 
-        self.history = None
+        self.history = []
         self.LOG_NAME = None
         self.next_plot_episode = self.PLOT_FREQ
         self.init_plot()
@@ -110,15 +110,15 @@ class Plotter(rclpy.Node):
                 data[item.key] = float(item.value)
 
         if data["episode"] is not None:
-            if self.history is None:
+            if self.history is []:
                 self.history = [data]
             else:
                 self.history.append(data)
 
             if data["episode"] >= self.next_plot_episode:
                 self.plot()
                 self.next_plot_episode += self.PLOT_FREQ
-                self.history = None
+                self.history = []
         else:
             #rospy.logerr("Bad progress message.")
             self.get_logger().warn("Bad progress message.")
@@ -175,8 +175,8 @@ class Plotter(rclpy.Node):
 
         # update file
         #rospy.loginfo(f"Saving training progress to {self.LOG_DIR}{self.LOG_NAME}")
-        self.get_logger().info("Saving training progress to {self.LOG_DIR}{self.LOG_NAME}")
-        plt.savefig(self.LOG_DIR + self.LOG_NAME)
+        self.get_logger().info(f"Saving training progress to {self.LOG_DIR}{self.LOG_NAME}")
+        plt.savefig(self.LOG_DIR + str(self.LOG_NAME))
 
 if __name__ == "__main__":
     Plotter()

src/rktl_autonomy/nodes/rocket_league_agent

@@ -20,7 +20,7 @@ env = RocketLeagueInterface(eval=True)
 
 # load the model
 # weights = expanduser(rospy.get_param('~weights'))
-weights = expanduser(env.node.get_parameter('~weights'))
+weights = expanduser(env.node.get_parameter('~weights').get_parameter_value().string_value)
 model = PPO.load(weights)
 
 # evaluate in real-time

src/rktl_autonomy/rktl_autonomy/_ros_interface.py

@@ -12,7 +12,7 @@
 from threading import Condition
 import time, uuid, socket, os
 
-from gym import Env
+from gymnasium import Env
 
 import rclpy
 from rclpy.duration import Duration

src/rktl_autonomy/rktl_autonomy/rocket_league_interface.py

@@ -12,10 +12,9 @@
 
 # package
 from rktl_autonomy._ros_interface import ROSInterface
-from gym.spaces import Box, Discrete
+from gymnasium.spaces import Box, Discrete
 
 # ROS
-# import rospy
 import rclpy
 from rclpy import Node
 from rclpy.parameter import Parameter
@@ -25,7 +24,7 @@
 
 # System
 import numpy as np
-from tf.transformations import euler_from_quaternion
+from transformations import euler_from_quaternion
 from enum import IntEnum, unique, auto
 from math import pi, tan
 
@@ -143,7 +142,7 @@ def __init__(self, eval=False, launch_file=('rktl_autonomy', 'rocket_league_trai
         self.node = Node('rocket_league_interface')
         # Publishers
         # self._command_pub = rospy.Publisher('cars/car0/command', ControlCommand, queue_size=1)
-        self.node.create_publisher(ControlCommand, 'cars/car0/command', 1)
+        self._command_pub = self.node.create_publisher(ControlCommand, 'cars/car0/command', 1)
         # self._reset_srv = rospy.ServiceProxy('sim_reset', Empty)
         self._reset_srv = self.node.create_client(Empty, 'sim_reset')
 
@@ -256,14 +255,15 @@ def _get_state(self):
         goal_dist_sq = np.sum(np.square(ball[0:2] - np.array([self._FIELD_LENGTH/2, 0])))
         reward += self._GOAL_DISTANCE_REWARD * goal_dist_sq
 
-        if self._score != 0:
+        if self._score is not None and self._score != 0:
             done = True
             if self._score > 0:
                 reward += self._WIN_REWARD
             else:
                 reward += self._LOSS_REWARD
 
-        x, y, __, v, __ = self._car_odom
+        if self._car_odom is not None:
+            x, y, __, v, __ = self._car_odom
 
         if self._prev_vel is None:
             self._prev_vel = v

src/rktl_autonomy/scripts/eval_rocket_league.py

@@ -6,7 +6,7 @@
   All rights reserved.
 """
 
-from rktl_autonomy import RocketLeagueInterface
+from rktl_autonomy.rocket_league_interface import RocketLeagueInterface
 from stable_baselines3 import PPO
 from stable_baselines3.common.vec_env import SubprocVecEnv
 from stable_baselines3.common.env_util import make_vec_env

src/rktl_autonomy/scripts/train_rocket_league.py

@@ -6,7 +6,7 @@
   All rights reserved.
 """
 
-from rktl_autonomy import RocketLeagueInterface
+from rktl_autonomy.rocket_league_interface import RocketLeagueInterface
 from stable_baselines3 import PPO
 from stable_baselines3.common.vec_env import SubprocVecEnv
 from stable_baselines3.common.env_util import make_vec_env
@@ -39,7 +39,7 @@ def train(n_envs=24, n_saves=100, n_steps=240000000, env_number=0):
 
     # log model weights
     freq = n_steps / (n_saves * n_envs)
-    callback = CheckpointCallback(save_freq=freq, save_path=log_dir)
+    callback = CheckpointCallback(save_freq=int(freq), save_path=log_dir)
 
     # run training
     steps = n_steps

src/rktl_autonomy/scripts/tune_rocket_league.py

@@ -6,7 +6,7 @@
   All rights reserved.
 """
 
-from rktl_autonomy import RocketLeagueInterface
+from rktl_autonomy.rocket_league_interface import RocketLeagueInterface
 import numpy as np
 from stable_baselines3 import PPO