Why use lasers? Why not just monitor input power and output power, and if the difference is too great infer that the power is being lost somewhere (ie. being absorbed by an obstruction) and back the power off until the losses/obstruction go away? Either have a dedicated radio /other data link to send this information between the rx/tx, or the information could be modulated onto a power beam if there happens to be one running in that direction. Such a system would be inherently safe compared to defining a beam width with lasers, as it is directly monitoring the thing that causes damage (absorbed power).

(I guess this idea can no longer be patented if it is novel, as this comment has just established prior art!)