Interesting problem... the alignment issues give me some concern too. You might get around this somewhat by having a large lens to collect the light prior to the detector itself. You'll want to mount the mirrors to something solid and substantial. There will be a need to tweak the mirror rotation in two planes.
Any plans to modulate the laser in order to improve the ability to reject ambient IR?
for reference, there are industrial items of this sort produced. The industry term is "light curtain", usually involving a number of parallel light beams for the purpose of detecting if something crosses a boundary or gets near a machine.
Other thoughts that stray off topic... you could use a rotating laser radar (lidar) to measure the distance to everything nearby and determine if something has moved into the area. Something like the Velodyne LIDAR, perhaps...
http://velodynelidar.com/lidar/lidar.aspx
It's not cheap, but it is cool.
Other thought.. instead of a rotating laser, maybe you could check the distance to nearby objects with a single beam of light that shines up from the ground and hits a conical mirror surface such that it causes the light to propogate horizontally in all directions. Then you just have to detect the delay from the transmission of the light to the reception of the reflections, and you'll know the distance to the nearest object. Piece of cake, eh?