The Development of Laboratory Work on the Topic: Pre-processing of Information from CVS Sensors of a Mobile Rescue Robot in Smoke Conditions

Bauman Moscow State Technical University, 2nd Baumanskaya str., 5/1, 105005, Moscow, Russia

^* Corresponding author: rubtsov@mail.ru

Abstract

The solution to the problem of processing long-range and television information received by the sensors of a mobile rescue robot in a smoke-filled environment is considered. A selection of budget sensors is made among those available in the free sale and having open-source software. The selected sensors are linked into a single information field in the free ROS software package using open-source libraries. The first stage of processing is the calibration of sensors to reduce the effect of distortion, as well as comparing the color image of the television camera with the readings of the rangefinder. The second stage is the analysis of existing solutions for image filtering in smoke conditions and the selection of the best according to the criteria for reducing the number of “smoke-filled” pixels and speed of response. In this paper, an algorithm is selected based on an atmospheric physical model with image analysis in the YCrCb space. The operation of this algorithm is demonstrated and a method for approximating a long-range image using a filtered color image is proposed to restore information from a rangefinder and further construct a model of the environment. Suggestions were made for further analysis and improving the accuracy of the algorithm. Based on this decision, laboratory work was formed in the course “RS designing”.