How weights are used in RLowess algorithm

I have a question regarding RLowess. On this page under Robust Local Regression. In the third step, it says "Smooth the data again using the robust weights. The final smoothed value is calculated using both the local regression weight and the robust weight". I am a little bit confused here, does this mean after computing robust weights, first the algorithm calculates Y values using the robust weights, then again caluclates Y values using both the local regression weight and robust weight? Or just after computing robust weights, it uses both of the weights (the local regression weight and robust weight) to calculate Y values? Furthermore, how the two weights are combined together? Does it mean each Wi is the multiplication of its corresponding local regression weight and robust weight?