The Task Planning Of Lunar Rover Parking In Place To Probe Based On SA* Algorithm

The task planning of rover parking in place to probe is an important part of rover autonomous detection system. The purpose of the planning algorithm is that to satisfy the constraints that manipulator should not collide with the detection environment, number of switches of manipulator is minimized and the work space of manipulator covers the detection point. Besides, the detection point should also satisfy constraints of smoothness and lighting to ensure that the spectrometer works correctly. In this paper, the main research object is the 3-DOF manipulator of the CE3 rover. In the paper, the forward and inverse kinematics model of the manipulator is researched. Based on the research, the work\detection space of manipulator, simulation of detection environment and collision detection of the manipulator is also researched. The smoothness and other constraints are satisfied by the estimation of detection point and calculation of the attitude of rover.First of all, the coordinate of each rod of the CE3 rover's manipulator was created using the D-H method. According to the rotation vector relation of adjacent coordinate, the forward kinematics of the manipulator is solved. The calculation of inverse kinematics was simplified by the transfer of the position and the Orthogonal configuration of the manipulator. The correctness of the forward and inverse kinematics of the manipulator is verified by the experiment. The concepts of work and exploration space are introduced using the principle of the forward kinematics, which establishes the theoretic foundation. The simulation of the environment, the collision detection method based on the AABB bounding box and virtual sensor is presented to satisfy the collisionless constraint. The safety of the new planning algorithm can be guaranteed by the collision detection method.Secondly, An improved SA* algorithm is proposed based on the analysis of A* algorithm. The constraint of minimum switches of manipulator is satisfied by costing cuts the same direction of the rotation and merging the adjacent points. The safety of the manipulator movement is guaranteed by the collision detection when the leaf nodes is extended. The test results show that SA* algorithm could plan the fewest control commands.Then, using the variance of the adjacent points of the detection point to evaluate the flatness of the probe surface. The lighting constraint is evaluated by calculating the angle between the sun reflected light and the spectrometer axis. Focusing on the problem of detection point is not in the work space and need to adjust the attitude of the rover many times, a new method of position and orientation is presented by vector projection and collision detection. The task planning of the CE3 rover parking in place to probe algorithm is summarized, which improves the detection efficiency.Finally, a 3D rover simulation software system was developed based on VS2008 and Open Inventor graphics library. The software integrates a variety of features. Not only proved the correctness of the principles involved in this thesis, but also improve the user experience. A fast method to extend HLA Web services is presented for data exchange and resource sharing in multi system. This method decouple the system and the programming language, reduce development cost and enhance the flexibility of the system development.