Research On SLAM Algorithm Of Robot In Unknown Environment

Mobile robot and slam technology is an important content of artificial intelligence research.Robot exploring unknown environment is realized by sensors on the body.Meanwhile,slam technology can continuously locate and map.From monoslam to PTAM,dso-slam and orb-slam algorithms,visual slam is widely used.Visual SLAM algorithm can be divided into two main directions: direct method based SLAM algorithm and feature-based SLAM algorithm.In the SLAM algorithm implemented by direct method,it can achieve better localization effect without extracting feature points,and can construct semi dense map.In SLAM algorithm based on feature method,the camera motion is estimated by feature extraction and matching.The greatest advantage of this algorithm is that it can detect and reposition loops,which plays an important role in building a global consistent map.An improved orb algorithm is proposed when studying SLAM algorithm.Orb algorithm uses fast corner and brief descriptor,which does not have rotation and scale invariance.The improved algorithm makes orb features adapt to rotation and scale changes by building image pyramid and using gray centroid method.The key frame selection strategy is also improved.The camera rotation and displacement are also considered.When the camera motion has enough rotation and displacement,the subsequent key frame insertion operation will be carried out.The improved algorithm can significantly reduce the number of key frames generated and reduce the time spent in global Ba optimization.Finally,the test is carried out in the data set and real environment to evaluate the trajectory estimated by the two types of algorithms.At the same time,the effectiveness of the improved orb algorithm and the improved key frame selection strategy is verified,and the maps constructed by the two types of algorithms are also evaluated.Figure62;Table5;Reference 50...