Design and Implementation of Artificial Intelligence in DOTA Games
introduce
In DOTA games and all real-time strategy games, artificial intelligence (AI) is an indispensable tool. Novice players have a preliminary understanding of the rules, fighting methods and heroic characteristics of the whole game through fighting against AI. AI is also the player's only opponent when not connected to the internet.
Traditional AI: the combat mode is single, the response speed is slow, and the action instructions are dull, so it is impossible to better analyze the battlefield situation and combat situation. Low level makes it easy for players to recognize the designed instructions, which leads to easy killing and affects the entertainment of the game.
Innovative artificial intelligence: it simulates human thinking and has a high IQ. Novice players can gradually learn how to play games and improve their understanding of the game through fighting with AI, instead of being ravaged by high-end players because of direct confrontation with humans.
In addition, innovative AI is not only aimed at novice players, but also has the characteristics of high level, wide routines and strong adaptability. At the same time, it can make high-level players have fun from playing against AI and further enhance the user experience.
The core function of 1 Artificial intelligence algorithm
The core functions of AI include control center, inspection system and command center. The detection system is like the eyes and brain of artificial intelligence. AI obtains game data through inspection system and analyzes these data. The analysis results of the patrol system are transmitted to the command center, which converts them into command signals and transmits them to the control center. The concrete behavior that the control center converts the signal into AI [1]. The specific functions are as follows:
2. implementation 2. Core function of artificial intelligence
2. 1 patrol system
Suppose a character's current health is H, attack power is A, defense power is D, magic power is M, and four skills are A 1, A2, A3, A4, and the corresponding four skill intensities are P 1, P2, P3, P4, and skill cooldown is C 1, C2, C3, C4.
If it is known that the remaining cooling time of any skill An is Yn and the total cooling time is Zn, there must be a linear function fn that makes the skill cooling degree Cn:
Cn=fn(Yn，Zn)，Cn？ [0, 1]
Calculating the cooling degree of skills in high-end games largely reflects a player's game level. When the cooldown of a skill is around 0, it means that this skill has just entered the cooldown time, and some heroic representatives who mainly rely on skills have lost their fighting power. When the skill cooldown is about to reach 1, the hero will recover his fighting power, which is equal to 1 and instantly recover a lot of fighting power. And this recovery process is often unexpected. In this process of restoring combat effectiveness, it is a process of testing players' technical content, such as retreating, ordinary attacking, moving or connecting other strategic plans [2].
Then, for Any skill, the corresponding relationship between the skill intensity Pn of an and the cooling degree Cn and other parameters X is as follows:
Pn=fc(Cn，X)
2.2 Self-study function
AI's self-learning makes AI improve its level in actual combat. In order to realize this function, we change the traditional routine of programming AI directly, let AI know what skills and attributes it has, tell AI the results of various behaviors, and let AI calculate the most effective routine in the current situation. This design enables AI to perform well in complex actual game battles [3].
2.3 Command Center
The command center is a system function module that converts signals into actual actions.
For example, the command center receives one from the control center? Attack? Command, then the command center will issue an attack command to the hero and feed back the result of this command to the control center. If the hero is stunned, then this command will definitely not be executed, then a stunned signal will be fed back to the control center, and the control center will immediately recalculate the countermeasures.
2.4 Control Center
Each subsystem is connected into a complete AI system through the control center, which receives the data and analysis results of each sub-function and then sends instructions to the command center.
For example, in actual combat, a high-explosive mage with control skills first sees an enemy in front of him. Through the system of knowing yourself and yourself, AI knows that the target enemy has lower fighting capacity than itself and can be killed. Then AI began to think about the killing strategy. Through the self-learning system, AI calculates the optimal scheme: first approach the target enemy by walking, and then try to subdue it with control skills. During the control skill, AI attacks the target enemy normally, and when the control skill is about to end, AI releases a big move to kill it. The results calculated by the self-learning system are transmitted to the control center, and the control center sends instructions to the command center, so AI takes action.
3. Conclusion
The design and development of artificial intelligence has further improved the shortcomings of traditional AI in real-time strategy games such as DOTA. From the player's point of view, artificial intelligence is designed for all levels. The demand of players is the basis of game design, and the experience of players is the goal of game design. Only by ensuring the player's excellent game experience can the game have greater market competitiveness.