zerosansan
commented
1 year ago Hi, I don't have the exact file on my local machine but I believe this is the same file ( https://github.com/zerosansan/td3_ddpg_sac_dqn_qlearning_sarsa_mobile_robot_navigation/blob/master/turtlebot3_rl_sim/src/configs/turtlebot3_world.yaml ).
Concerning your error, try changing Line 93 of start_dqlearn_training.py to:
network_inputs = 7
In my implementation, I used 54 because I feed in 52 Laserscan distance inputs and 2 goal information inputs (heading to goal and distance to goal).
You could try my new repository ( https://github.com/zerosansan/td3_ddpg_sac_dqn_qlearning_sarsa_mobile_robot_navigation) which removes openai_ros dependencies. So, it should be easier to get it running, the DQN implementation here is the same as my other repository. There are six RL algorithms, including Q-learning, SARSA, DQN, TD3, DDPG, and SAC.
Regards, Hafiq Anas
On Sun, Jun 25, 2023 at 1:11 PM Gabriela Xolo @.***> wrote:
Can you show us the file "openai_ros/src/openai_ros/task_envs/turtlebot3/config/turtlebot3_world.yaml"? i have a problem with the network "ValueError: Input 0 of layer "sequential" is incompatible with the layer: expected shape=(None, 54), found shape=(None, 7)" in the file start_dqlearn_training.py.
— Reply to this email directly, view it on GitHub https://github.com/zerosansan/dqn_qlearning_sarsa_mobile_robot_navigation/issues/1, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADEBUNVOAC7SNROSJKDTDNTXM7CBRANCNFSM6AAAAAAZS5PZAM . You are receiving this because you are subscribed to this thread.Message ID: <zerosansan/dqn_qlearning_sarsa_mobile_robot_navigation/issues/1@ github.com>
Can you show us the file "openai_ros/src/openai_ros/task_envs/turtlebot3/config/turtlebot3_world.yaml"? i have a problem with the network "ValueError: Input 0 of layer "sequential" is incompatible with the layer: expected shape=(None, 54), found shape=(None, 7)" in the file start_dqlearn_training.py.