Study On Trajectory Planning Algorithms For Robots And Their Projective Implementations Under Virtual Environment

The study on control simulation for robots based on virtual reality technology is one of the hot spots and emphases in the international research field of robotics at present. With the help of virtual reality technology, the simulation environment can be constructed more realistically and the interactive operation for man-machine interface can be executed more availably. The trajectory planning for robots is always the traditional problem in the field of robotics. At the present time, there are many optimization algorithms for the trajectory planning. Although those algorithms have their characteristics and play the important roles in some practical instances, they also have some essential drawbacks and deficiencies. So, it is important to discuss the refinement algorithms for the sake of widening the field of industrial applications for them and getting over the lack of earlier approaches.The study in this dissertation is carried out around the symbiosis of virtual reality technology with versatile robot trajectory planning. By means of introducing the virtual reality technology into the field of robot trajectory planning, the gap between them is bridged effectively. Taking the improving and expanding for robot oriented projective virtual reality (PVR) technology as the core of the study in this dissertation, it is analyzed with emphases on the research and development of PVR-based virtual simulation control platform for robots, the designing for the system application interfaces, and the presenting of optimization algorithms for robot trajectory planning in the projective virtual environment. The main contributions can be summarized in the following five parts.1. After surveying the international research state of the correlative fields in this dissertation in its entirety, it is analyzed with emphases on the advanced PVR technology concerning the symbiosis of virtual reality technology with versatile robot control technology. Here, the further research issues of PVR technology are proposed. Meanwhile, the architecture of PVR-based robot virtual simulation control platform is presented and implemented for the first time. It overcomes some drawbacks appeared in the original applications of PVR. Especially, among the improvements, the collision detection technology, the interactive operation for man-machine interface, and some details are discussed. Compared with other simulation platforms, this virtual simulation platform has simpler architecture and more extensive application fields. Infact, firstly, it can be used as the virtual simulation training platform and the realistic teleoperation training can be carried out based on it. Secondly, it can be also used as the experiment platform for multifunctional robots.2. Task deduction (TD) and action planning (AP) are two of the key components of PVR technology. But, it is difficult to construct TD and AP in unstructured and less known environments. Here, in order to overcome that deficiency, event-condition-action (ECA) rules of active database (ADB) are introduced into PVR-based simulation system to present a new software framework named TD&AP_ECA for the first time. Framework TD&AP_ECA not only has favorable flexibility (namely, it can re-configure the PVR system according to operators' requirements), but also provides an excellent application interface for collaborative and distributed applications of PVR. The feasibility and availability of the proposed methods are demonstrated through a preliminary simulation case study.3. Here, as one of key functions on the proposed simulation platform, several trajectory planning algorithms for robots in the projective virtual environment are studied respectively. First of all, the algorithms involving the minimum time trajectory planning (MTTP) and integrated optimization of trajectory planning (IOTP) for robots are studied by and large. In the proposed algorithms, combining one kind of designed simple collision detection strategy and some special optimization techniques, optimization model is presented entirely. In addition, the novel intensified evolutionary programming algorithms are presented to solve the corresponding models. The analyses and experiments show that the feasibility and availability of the proposed approaches. Especially, their computational accuracy is superior to existing methods. Finally, using M0T0MAN-UP6 industrial robot as the application prototype, the presented algorithms are implemented on the PVR-based robot control platform for the first time and some useful metaphors used to enhance the precision of projective operability are gotten. Here, those metaphors are named "virtual trajectory" specially.4. In the following, the minimum time motion path planning (MTMPP) algorithms for robots are studied. To deal with the original MTMPP problem, a novel hybrid evolutionary computation simulated annealing algorithm EC SA is presented. Meanwhile, the mathematic prove for convergency of the proposed method is given. The advantages of that novel optimization algorithm are demonstrated through a comparison with existing preferable elastic net method (ENM) and some otheroptimization algorithms such as simple genetic algorithm (SGA) and simulated annealing (SA) algorithm. Then, combining EC*SA and some special optimization techniques in robotics, three new optimization algorithms are presented respectively used to successfully solve three classes of MTMPP problems in some industrial contexts. Here, in the same way, the presented algorithms are introduced into projective virtual environment on the PVR-based simulation platform for the first time and some useful metaphors are gotten availably. Especially, those metaphors are named "virtual instructional moving path" which can be also used to greatly enhance the precision of projective operability.5. Here, based on the ant colony optimization (ACO) algorithm which has been applied widely in all kinds of fields because of its excellent ability of optimization, the intensified ant colony optimization (IACO*) algorithm used to solve path planning for robots is presented for the first time. Indeed, algorithm IACO can solve the continuous function optimization problem containing constraint conditions which has not been discussed at present. Especially, in the proposed method, some important contributions have been done in a creative way, including construction of the heuristic distance information probability, dynamical determination of the important weight coefficients, construction of the feasible solutions with high efficiency, and selection of the transition probability for ants in the moving process. In addition, in view of the present status that there are few proves for the convergency of ACO algorithm, based on the functional analysis, the preliminary mathematic analysis and prove for algorithm IACO have been given. The simulation results validate the effectiveness of algorithm IACO*.In the last chapter, after summarizing the contents of this dissertation, the further research issues concerning two theoretical research directions and one application direction are presented in detail, which will guide the future research direction in this field. Finally, considering the practical situations in our country, the future comprehensive applications of research efforts made in this dissertation are described.