When talk about racing game, the idea of the racing game is simple which is stay on course and be the fastest to the finish line to be a winner. However, that concept can be twisted to get more exciting racing game. According to Nitschke, racing game can be categorised into multiple types which are realistic racers, street racers and fun racers (Nitschke, 2007). Realistic racers focusing on realism aspects like Gran Tourismo and Colin McRae. Realism aspects is not important in street racers but it’s more like watching cool movies that broad to other aspects like able to tuning cars or listening to cool music while racing on the tracks like Need for Speed Underground. Fun racers like Trackmania and Carmageddon are more to fantasy racing with crazy tracks and able to jump high. Furthermore, the mission is more on achieving goal in limited of time.
In developing content for racing game, there are 3 approaches that can be applied to computational intelligence in the game. There are optimisation approach, innovation approach and imitation approach (Togelius, Nardi, & Lucas, 2007). In optimisation approach, the optimisation algorithm is used to change values of some particular aspects to get good and realistic game strategies. To implement a racing game, developers have to make an experiment to gather racing data especially the car physics and “make sense” tracks that can be built. The data gathered then can be applied to build the realistic game. Therefore, the fun racers category is exclusive in this approach.
To achieve realism in racing game, several physics concepts can be applied to the game such as gravitation, speed and inertia, friction and car’s engine torque. For instance, to enhance a game visual realism and realistic vehicle performance, multi-body vehicle suspension theory can be applied which is composed of springs, dampers and linkage through the wheels stay connected. “The suspension is what allows the wheels to have some range of freedom to move in the vertical direction without instantly lifting the corners of the vehicle. This enables the wheels to move in relative to the surface bumps and transfer less force to the vehicle body absorbing most of the shock” (Sharif & Brindle, 2007). This concept is matter much in game simulation because it has direct effect to the acceleration, handling and breaking performance of a vehicle due to the torque generated at its centre of gravity, the body tends to roll and pitch while accelerating or cornering at high speed (Sharif & Brindle, 2007). Unfortunately, very few of racing games implement this kind of physic concept.
Innovation approach is focusing on generating interests. The street racers and fun racers are examples that inclusive in this approach. In other words, this approach sees games as environments for the development of complex intelligence rather than computational intelligence techniques as means of achieving particular results as games.
The imitation approach is based on numerous of supervised learning. Symbolically, what is imitated is a human player, but a game agent can possibly imitate another agent. A good example of this approach to computational intelligence in racing games is the Microsoft Game Studio Xbox game Forza Motorsport. In this game, the player can train his avatar to play just like himself, and then virtual copy of him is used to get ranked on tracks, or test his skill against other players’ avatar (Togelius et al., 2007). In fact, this approach is used in my project to generate a “virtual bot” to the game. This implementation will be discussed in the next post.