Simulation and analysis of terrain sensing, position determination and navigation by a robotic lunar rover

Abstract

To aid in the development of a lunar rover, a series of robotic simulations was created using ROBOSIM and MATLAB Simulink. The first of these simulations utilized an invisible Cartesian robot along with collision detection heuristics within ROBOSIM to create an algorithm that emulated the surface mapping of the Moon by a lunar reconnaissance orbiter (LRO). The surface map produced by the LRO scan was used with a second simulation where the rover under supervised control, and with the use of a laser range finder, measured its distance to prior mapped terrain features. The measured distances, along with the position coordinates of the mapped terrain features were used to calculate the rover’s position. The experimental results show that on average the rover’s position was determined to within 3.31 meters of its true position. These results lead to the conclusion that this method of position determination can provide accuracy similar to the Earth-based global positioning system (GPS). To further analyze the effect of this error as well as the drift associated with an inertial measurement unit (IMU), a dynamics and navigation simulation was created in MATLAB Simulink. It was concluded based upon the results of the dynamics and navigation simulation that the IMU errors degrade performance far more than the initial position errors. Based upon this conclusion, future efforts can be focused on the modeling and performance of the IMU.