A Study on Optimal Design of Image Traverse Networks for Mars Rover Localization

Rover localization accuracy is critical for Mars surface exploration missions. It is especially crucial for future longrange rover missions, such as MER (Mars Exploration Rover) 2003. The geometry of the rover image traverse network is one of the most important factors to achieving high accuracy. A systematic study of geometric image traverse design was carried out. Several traverses of side looking and forward-backward looking rover image networks were simulated. Factors that affect the rover localization accuracy were examined, including traverse leg length, convergence angles, tie point distribution, effectiveness of cross-station tie points, etc. Accuracy values under different configurations are computed using a least squares bundle adjustment. Image resolution and similarity between adjacent images are analyzed as considerations for the automation of rover localization. Based on experiment results, conclusions and suggestions are given for optimal rover traverse design. Experiment results using actual data from Earth and Mars are also briefly illustrated.