Simultaneous Localisation Andmapping For Mars Rovers

As Mars exploration becomes more and more important and popular for the mjor space powers,studying and developing rover systems applied to complex environments has being the key technology for Mars surface research. Simultaneous Localisation and Mapping(SLAM) is the process of concurrently building a feature based map of the environment and using this map to obtain estimates of the location of the rover. This algorithm has the ability that simultaneously computing an estimate of rover location and an estimate of landmark locations by relative observations of landmarks. While continuing in motion, the rover will build a complete map of landmarks, which could be used to provide continuous estimates of the rover location and a basis for the next steps such as avoiding obstacles. Because of the upper advantage, it is necessary to apply the SLAM algorithm to the rover system.In this paper, we will study the 3-dimension extended Kalman filter based SLAM using to Mars Rover systems, including the following contents:Firstly, the system model in 3d Martian surface environment of Mars Rover is built, consisting of the definition and conversion of each coordinate system, the 3d and six DOF Rover motion model, and the 3d laser range finder mode. Then, general Extended Kalman filter based SLAM problem is illustrated. The computational complexity and consistency theoretical analysis for EKF-SLAM is provided. A Mars terrain constraint motion trajectory building method is established, simulation and analysis for 3d EKF-SLAM problem is done by using this method.Secondly, Constrained Local Submap Filter is studied. It creates an independent, local submap using observations of the features present in the immediate vicinity of the Rover. This local submap is then periodically fused into the global map to recover the full global map estimate. This representation is shown to reduce the computational complexity of maintaining the global map estimates.Thirdly, the convergence properties and consistency of extended Kalman Filter based SLAM algorithms are investigated. Proofs of convergence are provided for the nonlinear 3d SLAM problem with point landmarks observed using a ranged-bearing-tilting sensor in different motion scenario. Using those convergence properties shows “Why and how inconsistency can occur in the nonlinear EKF-SLAM”.Finally, Rover body-fixed SLAM is provided, which using the reference frame attached to the Rover as base reference. Due to omit Rover location and orientation uncertainty accumulation, this approach improves linearization of the model equations, which lead to the inconsistency reducing. This method is also combined with Constrained Local Submap Filter to obtain a high consistency and low computational complexity. The result by simulation verify the effectiveness of algorithm proposed earlier.