search

reiniscimurs / DRL-robot-navigation

Deep Reinforcement Learning for mobile robot navigation in ROS Gazebo simulator. Using Twin Delayed Deep Deterministic Policy Gradient (TD3) neural network, a robot learns to navigate to a random goal point in a simulated environment while avoiding obstacles.
MIT License
486 stars 97 forks source link

About Car Overturning #116

Closed dianstudy closed 2 months ago

dianstudy commented 4 months ago

hi,I followed your steps for training. At the beginning, the training was normal, but after certain episodes, it turned and overturned. May I ask if there is any solution？

reiniscimurs commented 4 months ago

Hi,

Please provide detailed description of the issue. Preferably with images and/or videos.

dianstudy commented 4 months ago

Hi,

Please provide detailed description of the issue. Preferably with images and/or videos.

ok.like this video? https://github.com/reiniscimurs/DRL-robot-navigation/assets/99256528/535bd7ff-87b3-43a7-9107-ca2b56c9a2be

and this is training situation. IMG_20240223_163241.jpg

reiniscimurs commented 4 months ago

Most likely your robot is not spawning properly after a reset and its wheel doesn't spawn in time before the robot starts moving. You can try to confirm this visually by looking at robot wheels at the reset time.

You can also try increasing the wait time in the reset function so that robot stays still for longer after reset: https://github.com/reiniscimurs/DRL-robot-navigation/blob/101ed8eaba77b9f447e3d57cc71518cfedb7f65c/TD3/velodyne_env.py#L279