| Biped climbing robots refer to those special climbing robots configured multiple degrees-of-freedom body in the middle and attaching devices at both ends. They usually move by them-selves through interchanging grippers to attach and cooperating with the body motion such asflexion and extension, swinging-around or flipping-over and so on. Compared with traditionalwheeled or tracked climbing robots, biped climbing robots feature high mobility in terms ofmultiple gaits, superior ability to negotiate obstacles and to transit between poles, etc. Hencethey are expected to be the automated solution to carry out high-strength, high-danger andhigh-rise tasks in stead of human beings. However, as regards to the state of the art of bipedclimbing robots all over the world, they lack of the capability to plan a path on spatial trussesautonomously and intelligently.This dissertation focuses on the fundamental issues related to collision-free path planningof biped climbing robots enclosing a pole to grasp on spatial trusses. Specifically, the maincontributions are as follows.(1) The description of spatial trusses and biped climbing robots is investigated, whichleads to a novel method with double configuration matrixes to completely express the re-lationship between them. The universal mathematical model of collision-free path planningwith biped climbing robots on spatial trusses is proposed. Inspired by biped walking, a three-layered framework integrating three planners, which are respectively the global path planner,the footholds planner and the single-step collision-free motion planner, is presented to solvethis problem. The roles of the three planners are also illustrated.(2) In order to fast filter potential poles for transiting, an approach valuating the accessibil-ity of position and the safety of grasps is proposed. A strategy based on the closest points of twopoles is also proposed to select the grasping orientation for transiting. The optional graspingregions for transiting with a five degrees-of-freedom biped climbing robot like Climbot-5D arediscussed and solved. In the basis of the above theories and approaches, a global path planneris designed to search the truss, so as to obtain roadmaps with global guidance information. Al-gorithms to select the grasping orientation and to solve optional grasping regions for transitingare verified through simulations.(3) According to the information of three basic gaits for biped climbing robots, the re- lationship between the mobile distance and the least number of climbing cycles is discussed.An algorithm based on this function relationship is presented to achieve the least number ofgrasping points through the whole path. Then the concept of graspable regions is defined, anda novel and fast approach using the binary approximating principle is proposed to solve them.After that, an algorithm based on solving the graspable regions is also presented to implementfootholds planning. With the above two algorithms, a footholds planner is designed to extractfootholds sequence from the global roadmap. Simulations demonstrate the efectiveness andthe efciency of the proposed algorithm to solving the graspable regions.(4) The specific model of single-step collision-free motion planning with biped climbingrobots is presented. The bilateral rapidly-exploring random tree (BiRRT) algorithm is intro-duced to solve this problem by universal design of its main processes, including state definition,sampling and interpolation, and so on. A preprocess is proposed to select suitable initial andgoal configurations for BiRRT. The efectiveness of the algorithm is verified by simulations.(5) Three planners are integrated together and their cooperation mode is also discussed.The specific steps to perform climbing path planning with them are presented. Through simu-lations with biped climbing robots with five and six degrees of freedom, the accuracy and theefectiveness of the presented theories, strategies and algorithms are verified. |
PDF Full Text Request