Loading...

Optimal Maneuver-Based Motion Planning Over Terrain and Threats Using Heuristic Optimization Approaches

Karimi, Jalal | 2012

627 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 42611 (45)
  4. University: Sharif University of Technology
  5. Department: Aerospace Engineering
  6. Advisor(s): Pourtakdoost, Sayyed Hossein
  7. Abstract:
  8. In line with advances in science and technology in the areas of robotics, artificial intelligence, computer and control, unmanned vehicles are being more utilized and focused. On the other hand, development and further advances of these vehicles is highly dependent on the level of their autonomy. Motion planning is one of the most important issues in this regard. Due to high levels of autonomy required for unmanned air vehicles (UAV), the subject of motion planning is of valuable interest in aerospace applications. The goal of the motion planning problem, that is the subject of this research, is to extract an optimal trajectory for the UAV from an initial location toward its target point while considering the terrain and threats. The requirements of this problem make its solution complex and computation intensive. To overcome the inherent complexities associated with this problem, many researchers have sufficed using only the kinematic equations of motions and a few performance constraints. But, to consider full dynamics of the UAV and its constraints requires utilization of the 6DoF nonlinear sets of motion equations whose solution is obviously timely and computation intensive. On the other hand, in order to acquire feasible trajectories one needs to consider the vehicle full performance, maneuvering capabilities and limitations. The proposed maneuver-based motion planning approach is able to handle some of these difficulties. In this approach, the nonlinear and complex dynamic of the vehicle is converted to a set of finite motion primitives, in derivation of which, the vehicle full performance and dynamic peculiarities are considered. The current thesis proposes novel solutions for the motion planning problem through development of new heuristic optimization algorithms. In this regard, a trim-maneuver library is initially formed. Where, the trim states are determined via a new constrained optimization approach. In addition, the maneuver trajectories are derived using a nonlinear constrained backstepping approach. Subsequently, the motion planning problem is studied in two off-line as well as real-time fashions. For the off-line problem, the Particle Swarm Optimization (PSO) algorithm is enhanced to generate the optimal trajectories. For the real-time motion planning, first, a new dynamic heuristic optimization approach is developed, tested against various benchmark problems and compared with the existing heuristic optimization algorithms. Then, the dynamic optimization algorithm is next adapted for the real-time motion planning in which stochastic threats are considered within the terrain. Evaluation of the proposed algorithm against several simulated scenarios has effectively demonstrated its potential for generating optimal contour-matching trajectories that succeed in avoiding stochastic obstacles. In addition, a trajectory control system designed and implemented on the vehicle shows that it can properly keep the vehicle on the optimal, terrain-threat observed, generated trajectories
  9. Keywords:
  10. Motion Planning ; Real Time ; Particles Swarm Optimization (PSO) ; Dynamic Hybrid Optimization ; Maneuver Automaton ; Terrain Obstacles ; Threat Zone

 Digital Object List

 Bookmark

No TOC