Real-time laser beam alignment to a remote target using blob detection and filtration for unmanned mobile spectroscopy platform

In a project supported by NASA, a new unmanned air quality monitoring system has been proposed to look for signs of life on Mars. The system development and tests are specially directed to build an unmanned open-path spectrometer for near-ground biogenic gas concentration detection on Mars. Moved by a mobile robot platform, the spectrometer shoots an infra-red laser beam to a remote retro-reflector and examines the reflected signal looking for the absorption peak of the investigated gas. The problem is that the laser beam needs to be constantly aligned to the retro-reflect (the target) while the mobile robot is moving. Since the target's location is unknown in advance, the system has to automatically survey and detect it before the alignment process could be started. Furthermore, the alignment process has to be performed automatically since it is hard to be done remotely from the enormous distance to Mars. Even then, the system should constantly keep the alignment active since the mobile robot is moving.;This work investigates the automation of the proposed monitoring system. The goal is to develop a fully automated surveying system using a single monocular camera and a pan/tilt unit. The surveying system is controlled by an automatic target detection and alignment system (ATDAS) program to survey for a remote, passive, and stationary target without prior information about its location. Blob detection and filtration under OpenCV library has been used to recognize the target from various distances, view angles, and environmental conditions. When the target is detected, the ATDAS system performs fast and accurate laser beam alignment to the target and keeps it aligned in real-time while the mobile robot is moving. Two formulas have been developed to drive the pan/tilt unit and adapt its panning and tilting speeds to the current zooming level of the camera. Furthermore, two methods have been developed to automatically estimate the distance to the target viewed from any distance or 3D angle based on the parallax method and the pinhole camera model (PCM). Finally, a completely randomized factorial experiment has been designed to analyze the effect of various factors on the accuracy of the developed PCM method for distance estimation.