Visual SVSF-SLAM Algorithm Based on Adaptive Boundary Layer Width

Fethi Demim,Abdelkrim Nemra,Kahina Louadj,Abdelghani Boucheloukh,Elhaouari Kobzili,Ahmed Allam,Mustapha Hamerlain,Abdelouahab Bazoula
DOI: https://doi.org/10.1007/978-3-319-97816-1_8
2018-09-11
Abstract:This work presents a solution to the Simultaneous Localization and Mapping (SLAM) problem of Unmanned Ground Vehicle (UGV) supported by a stereovision camera. There are many ways to approach this problem, mostly based on the sequential probabilistic process. The EKF-SLAM algorithm is one popular solution based on Extended Kalman Filter (EKF) to solve this problem that struggles with Jacobians’ calculation. FAST-SLAM is another popular algorithm based on the Rao-Blackwellized particle filter that has issues with real-time implementation. As a means to ameliorate the SLAM solution limitations, especially when the process and observation models contain uncertain parameters, a new adaptive Boundary Layer With algorithm based on the Smooth Variable Structure Filter (ASVSF) is proposed to solve the UGV visual SLAM problem. Hence, the adaptive SVSF-SLAM algorithm is proposed with an original formulation. This algorithm was implemented on Pioneer 3-AT mobile robot, using a stereo vision sensor in 3D. Simulation results show efficiency and give an advantage face modeling uncertainties and noises and it has significantly improved the performance of the estimation process.
What problem does this paper attempt to address?