Research On Robot Motion Planning Method For Multi-constrained Manipulation Tasks

Efficient and reliable motion planning under various unstructured constraints is the basis and prerequisite for multi-degree-of-freedom robots to realize motion intelligence.In recent years,the problem of robot motion planning has received widespread attention,and a large amount of research work is devoted to improving the intelligence and autonomy of robots in unstructured constrained scenarios.However,from the perspective of various fields of production and life,robots,especially multi degree of freedom robots represented by tandem manipulators,have not yet achieved high autonomy and high reliability of their operations through existing planning methods.The main performance is the existing planning methods generally have problems of low planning efficiency and poor path quality for operation tasks with multiple constraints.The subsequent task execution failures and even the risks of human and machine safety make the motion planning technology unable to be used.Application and landing in actual scenarios.Therefore,researching a motion planning framework that can fully acquire and handle various complex constraints that may occur in the process of work,and has high efficiency and high reliability,which is an ideal way to solve the above problems,and will strongly promote human-machine collaboration research in many fields,such as physical interaction,imitation learning,and intelligent grasping,has a wide range of application prospects.This paper decomposes the various motion constraints that may exist in the manipulator's operation process into environmental constraints,task constraints and control constraints.They are respectively summarized,described and modelled,and on this basis,a series of control inputs are found to make the manipulator safe,reliably complete the scheduled tasks,to establish a set of efficient and stable manipulator motion planning algorithms suitable for various complex and coupled constraints.This article starts with the analysis and definition of the three types of motion constraints: environmental constraints,task constraints,and control constraints.Combined with their respective characteristics,the construction method of the constraint parameterized model is studied to provide a priori and input for the motion planning problem of the manipulator.Aiming at the problem that environmental constraints cannot be explicitly described in the high-dimensional configuration space of the robotic arm,which leads to the blind exploration process of the planning algorithm and low planning efficiency,the environment is studied based on the incremental Gaussian mixture model and the greedy maximization expectation estimation algorithm.Constraint probability model construction method,and enable it to have the ability to update online with environmental changes.Because of the variety of task constraints,strong nonlinearity and implicit definition characteristics,a task constraint model based on continuous action imitation learning is studied.The construction method uses hand-eye tracking,sliding window consistent sampling,motion clustering,and total variation noise reduction,so that the robotic arm can segment and learn multiple task constraints from a single continuous teaching action,and use manifold metrics learn to build an approximate model of task constraints,and further study the method of constructing a task constraint model library containing a variety of manipulated objects.Aiming at the low efficiency of the motion planning algorithm in solving the problem of operation planning with multiple constraints,this paper uses the prior knowledge of environmental constraints and task constraint models to study the motion planning methods of manipulators oriented to environmental constraints and task constraints.Aiming at the problem that the collision detection module seriously restricts the efficiency of the planning algorithm,based on the probability model of environmental constraints,a collision detection algorithm that takes into account reliability and high efficiency is studied,which improves the collision detection speed of high-dimensional sample points in the configuration space;The non-linear,zero-measure and unresolvable characteristics of the constrained manifold cause problems of low planning efficiency and poor reliability.Based on the prior information of the manifold approximation graph and the manifold metric in the task constraint model,the local optimal expansion and manifold projection and other methods guide and inspire the planning and search process.Moreover,combining with the collision detection module to construct an efficient and stable environment and task-constrained robot motion planning method.Aiming at the problem that the path quality generated by the motion planning algorithm is poor and the existing trajectory optimization algorithm is easy to destroy the already satisfied constraints,this paper studies a manipulator trajectory optimization method based on quadratic programming under linear constraints.Use gradient information to perform numerical optimization in the trajectory parameter space,use the collision backtracking strategy to convert environmental constraints into linear constraints and incrementally add them to the quadratic programming process,and project the optimized trajectory generated by each iteration to the task constraint manifold.Generate a final trajectory that fully meets environmental constraints,task constraints,and control constraints,so that the robotic arm can perform tasks with multiple constraints with optimal time efficiency,and propose a quantitative evaluation index for the quality of the robotic arm's trajectory for subsequent comparison and verification experiments Foundation.Utilizing the research results of planning problems with multiple constraints,a framework of robotic arm motion planning algorithms with higher efficiency and reliability is constructed,and simulation and experiment platforms under multiple scenarios are built.Starting with the environmental constraint and task constraint model,the validity and accuracy of the constraint model construction and update method are verified.Based on the constraint model,the motion planning experiments of the manipulator under various tasks such as no-task constraints,end attitude constraints,closed chain constraints and passive chain constraints are carried out,and comparison experiments with similar algorithms are carried out to prove the motion planning proposed in this paper.The high efficiency of the algorithm and the effectiveness of the constraint model.Based on the initial trajectory obtained by the motion planning algorithm,a series of trajectory optimization experiments were carried out using the trajectory optimization method oriented to control constraints,which proved that the modelling,planning and optimization algorithm framework proposed in this paper can efficiently and stably solve the environmental constraints and task constraints.Motion planning tasks with multiple constraints,including control constraints,lay a theoretical and practical foundation for the application of the framework in practical scenarios such as industrial production,logistics distribution,emergency search and rescue,and home services.