Some time ago, I made a game of Gomoku against personal planes. I think artificial intelligence is very interesting. So I have nothing to do recently, and I have thought about the artificial intelligence algorithm of robot soccer.
The main connection between robot soccer and Gomoku game is that you need to analyze the state of the entire board/course at any time and make the most appropriate response. The main difference is that the Gobang game is a turn-based game, while the robot soccer is an "instant" game, and the other five are As long as it is based on a certain piece of chess, robot soccer must consider 23 people/balls (chess).
The surface is quite different, so let’s analyze it specifically:
1) About the "instant" characteristics of robot soccer
22 people, plus a football, a total of 23 objects on the field are moving at the same time. We can't say that robot soccer is like a Gomoku. First, one person moves, others stand, and the first one moves and then the second person moves. . . . . This is not playing football, but performing shows. If you have a real-time strategy game like StarCraft in the past, when all the people are deployed, press a ready button to inform the computer, and then the computer slowly measures the tanks it controls. It is fun.
Someone will be able to say that multithreading can be used (I stated in advance that my understanding of multithreading is limited to a few simple routines, so the narrative may be wrong). That's right, multi-threading can really be processed synchronously, and for games like football, the players on the field are certain (regardless of being sent off by the red card), with 23 threads to control the person and the ball, it seems that there is nothing wrong with it. . But if it is "Starcraft", a new tank will be built to build a tank. To kill a soldier, a process will be destroyed. . . . . It seems that this overhead is relatively large. What's more, if the computer builds 100 tanks and adds 100 soldiers, the resources consumed will be considerable. Although I don't know how StarCraft implements real-time strategy, I think even with multi-threading, there must be a special algorithm to optimize it.
Is it so complicated? We will carefully analyze the relationship between real-time and turn-based systems. We will find that there is no essential difference between the two. More precisely, the turn-based system is actually a special case of real-time system! When I was playing Backgammon, I set some flags in order to synchronize with the computer. When the player plays the game, the flag is reset and the computer can play chess. After setting it off once, it can't continue, waiting for people to play chess. If we cancel these signs, then the computer will play chess according to the current board situation. If you catch up with its rhythm, you may win. If you can't catch up, just wait for the lose! This way, Gomoku becomes a real-time strategy game.
So I think that for robot soccer, only all the players controlled by the computer are required to rotate, so that each player can make the most correct response to the situation on the field. The speed of the computer is very fast. Although it is a round robin, it is hardly noticed in the order, giving the impression that it is being carried out at the same time. In fact, the multi-threading of the operating system is also achieved in this way. X86 architecture cpu, multi-threaded multi-tasking program is running in protected mode, the basic idea is to slice cpu time, this time on-chip processing a task, whether it is processed or not, the next time slice must be processed b task, one The loop loops and continues to perform the unfinished a task.
But there is still a problem. Since the computer's response is much faster than people's, if you design a human-computer interaction game like FIFA, it is very likely that the computer has calculated all the objects it controls many times, but people have not responded yet. For real-time games that emphasize responsiveness, this gap in thinking speed may be unacceptable. My opinion is that you can let the computer pause for a period of time after processing all the objects, such as 500ms, such as 1s (the specific length can be obtained after testing). Moreover, the length of the computer pause time can also be used to distinguish the difficulty. The lower the difficulty, the longer the computer pauses. Of course, the pause I am talking about here does not mean that the computer really stops processing anything. The computer still needs to calculate the new coordinates of each player according to the preset direction and speed of movement and continuously refresh to display the animation effect. Only every 1 second, special reactions (kicking, stealing, etc.) will be made to change the direction and speed of movement.
2) Robot soccer computer considers more points than Gobang
There are 23 objects on the field that need computer processing, and the situation seems to be more complicated than Gomoku. But do you really need to handle 23 objects indiscriminately?
Let's take a look at real football. I don't really like football, I don't know football rules and various tactics. I only talk about my feelings.
For the ball, although it is moving at any time, the abstraction seems to be very simple on the field: the movement in a certain direction, the speed of the movement, the coordinates currently in place, just three elements. Others are controlled by which party, and whoever kicks, does not need or should be considered (this is in line with oop thinking).
Let’s take a look at 22 athletes. In fact, we can find that players who are closest to the ball will only respond accordingly (including dribbling/kicking, passing, catching, stealing, shooting, etc., only players who are close to the ball will make or even defend At the time, it is often the player who defends the ball. In the end, it is very close to the ball. I haven't seen the ball in the frontcourt, but the players are fighting in the backcourt (of course, don't consider the small action:)). Therefore, players who are far from the ball need only run to the ball or even temporarily stop (this can be controlled by the random function. The influence of the random value is the player's enthusiasm and the distance from the ball. The active player runs more actively. Running closer to the ball is more active). Only players who are close to the ball will make special treatment. After such fine division, the overall algorithm idea is relatively clear.
Through the above analysis, let's summarize it now:
When designing a robot soccer game, you can use a round robin method to handle 22 players and 1 ball. In general, the processing method is: calculating the new coordinates by the original coordinates and the determined direction and speed. After calculating the new coordinates of all objects, refresh the screen to display the animation effect.
Every once in a while (or several times), the computer is allowed to make special operations (kicking, passing, etc.) on each object. The essence of these special operations is to change the speed and direction of 23 objects so that the objects in the next period of time The trajectory of the movement changes.
When doing special operations, you can traverse according to the player serial number. The things that need to be done for each player are:
1. Calculate how far this player is from the ball. If you exceed a certain distance, you can handle it at will (such as the above mentioned, use the random function to determine whether the person is running or stopping)
2. If you are close to the ball, check the players around the ball that are closer to the ball (considering the complexity of the time, it is not recommended to actually traverse the entire stadium, but to maintain a table to record each player's current and Ball distance)
3. If there is a player closer to the ball, the options are: closer, steal, etc. (I don't know much about football, I don't know what else to do). Random numbers can be used to determine the action to perform
4. If there is a player closer to the ball, you can conclude that the person is taking the ball. The optional operation is: move to him, prepare to move, and move to the opponent's goal to prepare for the ball.
5. If there are no other players closer to the ball, then it can be concluded that the player is taking the ball. At this point, you need to check whether there are any opponents in the surrounding safety distance, whether there are any players, how far away from the goal, etc., and make the pass, the shot, the tape, etc. (no need to really calculate the distance between others and themselves) Because it is close to the ball at this time, it can be neglected. It can be represented by the distance of other people from the ball. The specific principle, the kick ball will know :)). These operations are relatively simple, just use some selection statements. When there are multiple operations to choose from, you can use random numbers to determine. The various attributes of the player can affect the value of the random number.
In addition, there are some exceptions: for example, when there are athletes running, running, colliding, hitting the ball, etc., even if they have not routinely handled special operations, they must be targeted immediately. The players involved perform special operations, otherwise they are untrue and unreasonable.
Finally, let's talk about the data structure of players and balls.
Player:
{
Direction / speed / current coordinates. . . . .
Pass () / kick () / shot () / catch () / steal (). . . . . . ./Running ()
Calculate new coordinates ()
}
football:
{
Direction / speed / current coordinates. . . . .
Calculate new coordinates ()
Friend:
Player: Pass () / kick (). . . . . . //Or do an interface that changes direction/speed so that the function in the player can call this interface
}
In addition, some tables are needed to maintain information such as the distance of each player from the ball to facilitate various processing. There are also some details that may be necessary to make the game, but because it is relatively simple, there is not much analysis here.
Best Rechargeable Vacuum Cleaner,Rechargeable Hand Vacuum Cleaner,Small Rechargeable Vacuum Cleaner,Cordless Rechargeable Vacuum Cleaner
Ningbo ATAP Electric Appliance Co.,Ltd , https://www.atap-airfryer.com