Development of a Mars Exploration Rover for RASC-AL Exploration RoboOps Competition with an Extended Kalman Filter Based Navigation System.

摘要

The primary objective of this research is to develop a navigation system for use on a rover designed for the NASA-NIA Revolutionary Aerospace Systems Concepts Academic Linkage (RASC-AL) Exploration RoboOps challenge. This competition takes place at NASA's Johnson Space Center Rock Yard, a test facility with many obstacles including steep hills, craters, rocks and loose sandy terrain. In addition to the challenges of the terrain, the rover's operators must control it from West Virginia University's campus. It was observed during the 2012 competition that a primary challenge was a lack of situational awareness; the operators had to navigate the large test area with only delayed video from two low resolution cameras. This research seeks to provide better awareness of the rover's position in the rock yard through the development of an accurate navigation assistance system.;This research first presents the rover that was designed, developed, and demonstrated at the 2012 competition that features six-wheel-drive, four-wheel steering, and a rocker-bogie suspension system similar to NASA rovers currently operating on Mars.;This research then presents a Navigation Assistance System (NAS) capable of providing nine-state navigation data, using small, lightweight, and low-cost sensors. In order to achieve an optimal estimate of the rover's attitude, velocity, and position this system employs an Extended Kalman Filter (EKF) that fuses data from an Inertial Measurement Unit (IMU) and a Global Positioning System (GPS). The EKF utilizes complimentary data from both systems to minimize the degradation caused by inherent system error that would be experienced by using either system alone.;The primary output of the NAS, position, is evaluated against a high-quality dual-antenna, Differential GPS (DGPS) receiver, as well as a low-cost single antenna GPS receiver. The NAS was observed to provide smoother position output, however actual error from the DGPS was only slightly reduced; in the nominal case an improvement of 0.2 meters was observed. Furthermore, testing demonstrated that the overall accuracy of the GPS alone, in comparison with the DGPS unit, was observed to be 1.7 meters. For the intended purposes of the system, it was found that using GPS alone would be sufficient for navigation applications.;The integration of this system with the rover brought unexpected challenges stemming from electrical interference created by on-board electronics, therefore this work presents the results of the navigation assistance system development taken off-board the rover. Furthermore, it presents research on autonomous navigation algorithms that could implement the state estimation data provided by the navigation assistance system that was developed.