We have implemented a non-linear polynomial regression model on the Training-set to extract a parameter that will be a result of learning step. We will use this parameter for estimation which will be discussed in the next step. A non-linear polynomial regressor can have a degree 2 or higher. Initially, we have tried implementing with degree-2. Further to better fit the data, we checked with higher order degrees. Note that the increase in the degree of non-linearity will increase the size of feature-vector and hence the computational complexity. We have found that after a degree 3, a negligible increase in estimation accuracy was observed. Hence we have set a degree of non-linearity of 3 for our model. This resulted in the final feature-vector size growing to 90 for 30 features extracted. For each data-set, a 90 dimensional feature-vector x is extracted along with a 2 dimensional world-state vector w (ground-truth distance and intensity) as shown in Equation (1)