This dissertation presents a novel method to improve the chassis control performance of ground vehicles using understeer characteristics. It is well-known that most production vehicles are generally built to show some understeer characteristics for preventing the oversteer. That is, the brake systems are designed to be front biased such that the front wheel slip ratio are saturated first. Therefore, front wheel slip tend to be larger than the rear ones during braking. Similarly, the wheel sideslip angles at the front wheels should always have a larger value that those for the rear wheels during normal cornering. These braking and handling characteristics are exploited to classify the road surface condition in real-time. It is widely known that the tire-road friction coefficient is crucial factors for vehicle safety control systems. Therefore, numerous efforts have been made to resolve these problems, but none have presented satisfactory results in all cases. In this dissertation, cost-effective observers for estimating road surface condition are introduced first. The estimation processes are activated during braking or cornering, which is a quite reasonable. Unlike the conventional methods which concentrated on individual wheel’s state variables, i.e., state variables of front wheels, the propose method compares the front and rear wheel slip ratio and sideslip angles simultaneously. In this way, estimation of unmeasurable vehicle state variables such as wheel sideslip angle, slip ratio, and absolute vehicle velocity, which are usually unavailable in production vehicles, can be removed from the entire algorithm. This is the main contribution of the estimation part of this dissertation. Moreover, individual brake torque, which are usually assumed as constant values in previous studies, are also estimated in real-time by exploiting the understeer characteristics. By obtaining the accurate brake torque, the estimation performance of tire-road friction coefficient during braking can be improved. The all developed algorithms for estimating road surface condition are verified through simulations and experiments using a production vehicles, and it confirms that the estimation performance is improved significantly compared with those for the conventional methods. Based on the estimation results, the pure longitudinal and pure lateral control for improving the vehicle dynamic performance and vehicle safety are presented. Using a special type of sliding mode control, a new type of ABS and TCS for any types of ground vehicles are developed. In order to realize a robust wheel slip control system, the wheel slip and the desired wheel slip are generally required. Unfortunately, however, these parameters cannot be accurately measured in production vehicles. The method suggested in this dissertation is aimed at solving these problems by exploiting the nonlinear characteristics of tire force. The front and rear wheels have different roles in the proposed method. The in-wheel motors are controlled to cycle near the optimal slip point. Based on the cycling patterns of the front wheels, the desired wheel speed for rear wheel is defined. The rear wheels are controlled to track this defined speed. Regarding the vehicle lateral control, a novel yaw rate–based lateral control in that the available road surface friction is fully utilized as much as possible is presented. The desired yaw rate is not explicitly defined in this dissertation. Instead, the difference of wheel sideslip angles between the front and rear wheels is regulated by the additional yaw moment. As a result, the vehicle is steered to a neutral steering condition in which the rear wheel sideslip angle increases up to the magnitude of front wheel sideslip angle. Thanks to this approach, the increased lateral acceleration compared to that of conventional methods can be realized, meaning the improved cornering performance. Moreover, the complicated estimation processes such as road surface friction, sideslip angle, and model parameters are excluded in the proposed method. As a result, the entire algorithm is much simpler compared with the conventional controllers, resulting in less computation. The developed methods are also confirmed in simulation and real car-based experiments, and the results reveal that the proposed method opens up opportunities for new types of chassis control systems using the characteristics of the motor.