Artificial intelligence for driving lines in racing games
Dynamic driving aides, such as driving lines, turn indicators, braking indicators and acceleration indicators, for example, can be provided for players participating in a racing game. Typically, driving lines are provided for each class of cars. However, even within a class of cars, each car differs enough that the ideal driving lines and breaking points can vary. Therefore, with an agent trained via reinforcement learning, an ideal lines and other driving aides can be established for every individual car. These guides can even be varied to account for variations in the weather or other track conditions.
1. A method for providing driving aides in a racing game comprising:
training a neural network for a trained agent to achieve optimal racing results in a particular car on a particular track;
operating the trained agent in parallel with a player racing in the racing game, where the trained agent is provided state information of the player during the racing game;
looking ahead, by the trained agent, to determine a recommended course of action based on the state information of the player, the recommended course of action determined by the trained agent to achieve the optimal racing result; and
providing course overlay information based on the recommended course of action.
2. The method of claim 1 , further comprising displaying the course overlay information on a player's display during racing by the player of the racing game using the particular car on the particular track.
3. The method of claim 1 , wherein method is repeated, changing the particular car or the particular track, to train the trained agent on each combination of possible cars and tracks in the racing game.
4. The method of claim 1 , wherein the course overlay information is stored within the racing game.
5. The method of claim 1 , wherein the course overlay information is stored in a cloud-based storage and is retrieved prior to starting a race in the racing game.
6. The method of claim 1 , wherein the course overlay information is based on in-game weather conditions during the player racing in the racing game.
7. The method of claim 1 , wherein the course overlay information is based additionally on a player's car velocity and current position in the racing game.
8. The method of claim 7 , wherein the course overlay information is dynamically computed during the player racing in the racing game.
9. The method of claim 1 , wherein the state information of the player includes at least one of a current a velocity of a race car of the player, a current trajectory of the race car of the player, attributes of the race car of the player and in-game weather conditions.
10. The method of claim 1 , wherein the course overlay information includes at least one of driving lines, braking suggestions, acceleration information, and a trajectory target.
11. The method of claim 10 , wherein the braking suggestions are provided as a function of attributes of the particular car.
12. The method of claim 10 , wherein the braking suggestions are provided as a function of a velocity of the particular car during the racing game.
13. The method of claim 10 , wherein the braking suggestions are provided as a function of a type of tire and an amount of wear of the tire.
14. The method of claim 10 , wherein the trained neural network is trained to provide the trajectory target based on adjacent or upcoming traffic encountered by a player during the racing game.
15. The method of claim 1 , wherein the trained neural network is trained to provide the recommended course of action based on adjacent or upcoming traffic encountered by the player during the racing game.