search

rajkhatik24 / Reinforcement-learning-based-approach-for-autonomous-driving-in-CARLA-simulator

The main goal of the project is to test Deep Reinforcement Learning based algorithm for autonomous driving application. CARLA simulator was used to evaluate the performance of DQN and DDQN model.
1 stars 0 forks source link

The results are not good #2

Open xss1006 opened 1 month ago

xss1006 commented 1 month ago

Hello, I tried to replicate your code, but I couldn't achieve the same results as in your demo. I used the exact parameters you provided. Do you know how to resolve this issue?

rajkhatik24 commented 1 month ago

Hi, If you got the code working and can see some movements where the autonomously tries to change its direction of driving, then all you could do is play around more with the hyperparameters. I believe since this is realtime training based on simulation, the results for the model may vary from system to system. Hope this helps.

xss1006 commented 1 month ago

嗨，如果你的代码运行正常，并且可以看到一些自动驾驶尝试改变行驶方向的动作，那么你所能做的就是更多地尝试超参数。我相信，因为这是基于模拟的实时训练，模型的结果可能因系统而异。希望这对你有所帮助。

Thank you for your reply. He keeps turning in circles, and I'm not sure if this is a normal phenomenon.