Journal of Shanghai Jiaotong University ›› 2018, Vol. 52 ›› Issue (1): 26-32.doi: DOI: 10.16183/j.cnki.jsjtu.2018.01.005

Previous Articles     Next Articles

3D Map Reconstruction in Indoor Environment Based on Normal Distribution Transformation Under Gravity Constraint

QI Mingxua,b,YANG Minga,b,WANG Chunxiangc,WANG Binga,b   

  1. a. Department of Automation; b. Shanghai Key Laboratory of Navigation and Location Services; c. Research Institute of Robotics, Shanghai Jiao Tong University, Shanghai 200240, China
  • Online:2018-01-01 Published:2018-01-01

Abstract: A method for offline reconstruction of indoor 3D map is proposed based on an improved normal distribution transform (NDT) algorithm. First, a constraint in the gravity direction is incorporated into a 3D NDT, so that map deformation in extra dimensions is rectified. Then a cyclic gradient matching scheme is proposed to deal with the matching error resulted from scarce overlap between consecutive frames. Finally, a graph optimization is adopted in the back-end map construction. By using a swinging laser, experiments validate its performance and high accuracy.

Key words: indoor 3D map, reconstruction, gravity constraint, normal distributions transform (NDT), cyclic gradient matching, graph optimization

CLC Number: