Generalized Single-Vehicle-Based Graph Reinforcement Learning for Decision-Making in Autonomous Driving
In the autonomous driving process, the decision-making system is mainly used to provide macro-control instructions based on the information captured by the sensing system. Learning-based algorithms have apparent advantages in information processing and understanding for an increasingly complex drivi...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-06-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/22/13/4935 |
| _version_ | 1797441872615964672 |
|---|---|
| author | Fan Yang Xueyuan Li Qi Liu Zirui Li Xin Gao |
| author_facet | Fan Yang Xueyuan Li Qi Liu Zirui Li Xin Gao |
| author_sort | Fan Yang |
| collection | DOAJ |
| description | In the autonomous driving process, the decision-making system is mainly used to provide macro-control instructions based on the information captured by the sensing system. Learning-based algorithms have apparent advantages in information processing and understanding for an increasingly complex driving environment. To incorporate the interactive information between agents in the environment into the decision-making process, this paper proposes a generalized single-vehicle-based graph neural network reinforcement learning algorithm (SGRL algorithm). The SGRL algorithm introduces graph convolution into the traditional deep neural network (DQN) algorithm, adopts the training method for a single agent, designs a more explicit incentive reward function, and significantly improves the dimension of the action space. The SGRL algorithm is compared with the traditional DQN algorithm (NGRL) and the multi-agent training algorithm (MGRL) in the highway ramp scenario. Results show that the SGRL algorithm has outstanding advantages in network convergence, decision-making effect, and training efficiency. |
| first_indexed | 2024-03-09T12:32:26Z |
| format | Article |
| id | doaj.art-0a168d3851f148cabd01119ad7588149 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-09T12:32:26Z |
| publishDate | 2022-06-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-0a168d3851f148cabd01119ad75881492023-11-30T22:27:38ZengMDPI AGSensors1424-82202022-06-012213493510.3390/s22134935Generalized Single-Vehicle-Based Graph Reinforcement Learning for Decision-Making in Autonomous DrivingFan Yang0Xueyuan Li1Qi Liu2Zirui Li3Xin Gao4School of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, ChinaIn the autonomous driving process, the decision-making system is mainly used to provide macro-control instructions based on the information captured by the sensing system. Learning-based algorithms have apparent advantages in information processing and understanding for an increasingly complex driving environment. To incorporate the interactive information between agents in the environment into the decision-making process, this paper proposes a generalized single-vehicle-based graph neural network reinforcement learning algorithm (SGRL algorithm). The SGRL algorithm introduces graph convolution into the traditional deep neural network (DQN) algorithm, adopts the training method for a single agent, designs a more explicit incentive reward function, and significantly improves the dimension of the action space. The SGRL algorithm is compared with the traditional DQN algorithm (NGRL) and the multi-agent training algorithm (MGRL) in the highway ramp scenario. Results show that the SGRL algorithm has outstanding advantages in network convergence, decision-making effect, and training efficiency.https://www.mdpi.com/1424-8220/22/13/4935autonomous drivingdecision-makinggraph convolutiondeep reinforcement learning |
| spellingShingle | Fan Yang Xueyuan Li Qi Liu Zirui Li Xin Gao Generalized Single-Vehicle-Based Graph Reinforcement Learning for Decision-Making in Autonomous Driving Sensors autonomous driving decision-making graph convolution deep reinforcement learning |
| title | Generalized Single-Vehicle-Based Graph Reinforcement Learning for Decision-Making in Autonomous Driving |
| title_full | Generalized Single-Vehicle-Based Graph Reinforcement Learning for Decision-Making in Autonomous Driving |
| title_fullStr | Generalized Single-Vehicle-Based Graph Reinforcement Learning for Decision-Making in Autonomous Driving |
| title_full_unstemmed | Generalized Single-Vehicle-Based Graph Reinforcement Learning for Decision-Making in Autonomous Driving |
| title_short | Generalized Single-Vehicle-Based Graph Reinforcement Learning for Decision-Making in Autonomous Driving |
| title_sort | generalized single vehicle based graph reinforcement learning for decision making in autonomous driving |
| topic | autonomous driving decision-making graph convolution deep reinforcement learning |
| url | https://www.mdpi.com/1424-8220/22/13/4935 |
| work_keys_str_mv | AT fanyang generalizedsinglevehiclebasedgraphreinforcementlearningfordecisionmakinginautonomousdriving AT xueyuanli generalizedsinglevehiclebasedgraphreinforcementlearningfordecisionmakinginautonomousdriving AT qiliu generalizedsinglevehiclebasedgraphreinforcementlearningfordecisionmakinginautonomousdriving AT ziruili generalizedsinglevehiclebasedgraphreinforcementlearningfordecisionmakinginautonomousdriving AT xingao generalizedsinglevehiclebasedgraphreinforcementlearningfordecisionmakinginautonomousdriving |