Environment Modeling Method for Intelligent Robots Based on Binocular Vision

Abstract With the development of technology, demand for intelligent mobile robots is growing steadily. Environment perception and modeling, are key elements to achieve robot intelligence. Therefore, binocular stereo vision sensor used as core device perform operations such camera calibration, obtaining internal/external parameters, distortion correction. On this basis, a multi-dimensional all-round environment system constructed, dynamic target detection results 3D scene map reconstruction displayed dynamically in real time. The experimental calibration show that re-projection error lies sub-pixel interval, image caused by basically eliminated meet expectation correction; modeling mean value deviation between object representation actual -0.041m, among which minimum 0.001m maximum 0.532m, excellent tasks feature point matching. In addition, storage space significantly reduced, real-time performance better than traditional method.