I am working on an autonomous driving domain and my field poses multiple autonomous vehicle collision avoidances. I want to try to solve it using RL. In this case, the following statements regarding the RL method are correct or any thins complicated with the fundamentals of RL. Please explain to me.
"Through Learning Control, control knowledge of a control function can be created through the training by Reinforcement Learning. However, the conventional Reinforcement Learning method does not provide the application of more than one control function within a Learning Control System. Execution of more control functions within a Learning Control System would require the application of multiple learning processes within a control system. Methods concerning the application of learning processes in Learning Control vary depending on the application of the control device and the purpose of the system."
Dear Abu Jafar Md Muzahid
Basically, Reinforcement learning (RL) is an area of machine learning concerned with how intelligent ... In the operations research and control literature, reinforcement learning is and intact maximizes the "reward function" or other user-provided reinforcement signal.
However, this articles might be helpful, have a look
1. https://www.researchgate.net/profile/Mohamed-Mourad-Lafifi/post/Reinforcement-Learning-in-Control-Theory/attachment/5f32faf16a5a03000178aa00/AS%3A923438789193728%401597176372063/download/Reinforcement+Learning+for+Contro+_+Performance%2C+Stability%2C+and+Deep+Approximators.pdf
2. https://cran.r-project.org/web/packages/ReinforcementLearning/vignettes/ReinforcementLearning.html
Kind Regards
Qamar Ul Islma
The problem under discussion is obstacle avoidance by reinforcement learning. Reinforcement learning works based on the concept of reward based action. So, every time a obstacle is detected, your logic should be such that advancement or activation happens on a particular front which could be treated as the reward. Hope this helps a bit.
- Badges
- Science topic