Slip-compensated odometry based on the slip estimation method of a tracked vehicle is proposed to improve the accuracy of position estimation. While a robot travels on a loose and weak slope, it slips longitudinally and laterally, particularly when rotating. These slips introduce error in odometry, and it is difficult to use odometry in controlling a robot and recognition of an environment such as trajectory following and three-dimensional mapping. In this paper, we describe a method of estimating these slips with inertial information, rotation velocity of the tracks employing a simple regression function. We confirmed that previous work of estimating longitudinal slippage on a horizontal plane can be applied for a weak slope. Estimation of lateral slippage is based on a regression analysis using offline training data of a robot's travel on a slope. This regression technique can be used online using information from a global navigation satellite system or other sensors. We applied these slip estimation methods to a kinematic model of a skid-steering tracked vehicle for odometry, and confirmed the improvement in precision of odometry compared with conventional odometry by conducting indoor sandy-slope experiments. We also proposed a path following control law considering lateral slippage, and confirmed the effectiveness of the control law in experiments.