A Motion Constraint Dynamic Path Planning Algorithm for Multi-Agent Simulations

T. R. Wan, H. Chen and R.A. Earnshaw

In this paper, we present a novel motion-orientated path planning algorithm for real-time navigation of moving agents. The algorithm works well in dynamical and un-configured environments, and is able to produce a collision-free, time-optimal motion trajectory in order to find a motion-optimised path. In addition to the motion modelling based path planning, our approach can deal with the obstacle-space unknown or partly unknown to moving agents. It therefore solves the drawbacks of traditional obstacle-space configuration methods. Multi-agents behaviour has been explored based on the algorithm. In the simulation a simple physically-base aircraft model has been developed, which is addressing the manoeuvring capabilities of the moving agents, while the moving agents' accelerations and velocities are always continuous and bounded. The generated motion path is constituted smoothly and has continuous curvature on the whole state space of the motion thus satisfying the major requirements for the implementation of such strategies on real-time animation or in simulation applications in VR environments.