This paper addresses semi-autonomous collision avoidance flight using two dimensional laser range finder with mirrors. One of practical applications of unmanned aerial vehicles (UAV s) can be found in inspection of deteriorated infrastructures such as bridges and tunnels. In such applications, proper environmental recognition and collision avoidance system are of importance because of GPS-denied environment. This paper develops environmental recognition and collision avoidance systems with less computational consumption using two dimensional laser range finder with mirrors. The sensor used in this study is able to scan over horizontal direction and measure vertical direction by changing the laser direction with mirrors. Consequently, the sensor can obtain three dimensional environmental information. In this paper, we describe altitude estimation method under rough ceiling of bridge and obstacle detection system in the flight direction from the point data. The fundamental flight experiments are carried out to demonstrate the validity of the developed system.