Autonomous Agent Path Finding

Transcript

1. Autonomous Agent Path Finding
2. Q-learning • • •
3. Deep Q-learning
4. Building Steps ▰ ▰ ▰ ▰ ▰ ▰
5. Modeling the environment. • • • •
6. Goal • • • •
7. Goal(cont.)
8. Setting parameters angle: rotation: pos=(self.car.x, self.car.y): velocity=(velocity_x, velocity_y): sensor1:(sensor1_x, sensor1_y): sensor2:(sensor2_x, sensor2_y): sensor3:(sensor3_x, sensor3_y): signal1: signal2: signal3:
9. • • sand • sand Setting parameters(cont.) Figure 2: Sand array get only 1’s in its first rows, remaining element get 0
10. Setting parameters(cont.) goal_x: goal_y: xx=(goal_x-self.car.x): yy=(goal_y-self.car.y): orientation:
11. • • orientation The input states
12. • • sensors • • The input states(cont.)
13. • • • • The output actions
14. • • The output actions(cont.)
15. • • • • The rewards
16. • • • The rewards(cont.)
17. REFERENCES ▰ ▰
18. THANKS! Any questions? ▰ ▰