The vertical load and lateral force parameters of sliding steering wheeled robot tires vary greatly and are difficult to estimate in real time due to the influence of orchard road surface undulations and tire adhesion capacity changes. The existing sliding steering controller is designed to simplify the tire dynamics parameters, which leads to the problem of low stability of robot attitude control. A real-time estimation method for the vertical load of sliding steering wheeled robot tires on unpaved roads and an algorithm for real-time estimation and optimal distribution of tire driving force were proposed. Firstly, an ideal plane for the static calculation of the sliding steering process and a four-wheel vertical load estimation method based on this plane were proposed; secondly, a small lateral deflection angle lateral force estimation method based on the Fiala wheel-tire dynamics model was proposed; and the steady-state dynamics equations of the sliding steering wheeled robot ramp and the real-time tire drive force estimation method were established; finally, the optimal real-time distribution model of the driving force was constructed based on the tire utilization rate. In order to verify the method, an ADAMS-based sliding steering wheeled robot dynamics model was developed for comparison and a testing device was built to verify the vertical load and lateral force estimation methods. The results showed that the accuracy of the real-time tire vertical load estimation method was more than 95% and the accuracy of the real-time lateral force estimation method was more than 85%, and the tire drive force optimization method based on the tire vertical load and lateral force reduced the tire utilization rate from 96.25% to 93.75%, which improved the tire adhesion margin and attitude control stability.