The Research On Environment Modeling And Motion Planning For UAVs

This paper takes a research on the algorithms about environment modeling and motion planning for Unmanned Aerial Vehicles (UAVs). It first models the global environment using the Dipole Ellipse Envelope (DEE) model and proposes the concept of the entropy production based on this model. This model consists of two modes, called the Isolated Envelope Mode and the Combined Envelope Mode, respectively. The former is designed for the isolated simple obstacles and the latter is suitable for the clustered complex obstacles. However, they hold the same philosophy, that is, trying to comprehend the outline and the trend of obstacles, instead of concentrating on the nitty-gritty along their borders, so as to make it more concise and flexible at some cost of the accuracy. This model is especially pertinent to the environment with unknown information or dynamic change.Based on this model, this paper distributes motion planning algorithms in a hierarchical structure:In the global environment, a multi-objective function is constructed by integrating the path length and the entropy production, and Particle Swarm Optimization (PSO) is applied to accomplish the path planning; while in the local environment, the environmental model is adjusted by sensing the surrounding environment in real time, and the Artificial Potential Field (APF) method is used to avoid collisions dynamically. In particular, the traditional APF method is improved in two aspects:first, a new component called the conductive potential field produced by the dipoles is introduced, which decreases the local minima as well as the trajectory oscillations; second, the relationship between the position and the velocity is considered for the mutual promotion between these two factors.To achieve the harmonious coordination, the path planning algorithm and the collision avoidance algorithm are integrated into a unified framework:the former is responsible for planning a rough path consisting of a series of intermediate nodes in the global environment, while in the local environment, the latter is used to avoid collisions with obstacles between two adjacent intermediate points. In such a framework, they promote themselves mutually by:on one hand, the global path is optimized by the former so that the latter will not suffer from the local minima as well as oscillations; on the other hand, the trajectory smoothness is improved by the latter so that the former can decrease the number of intermediate nodes, which makes it much faster.Based on the DEE model, the computer simulations are run on the path planning algorithm and the collision avoidance algorithm. It proves that they work fine in the unified framework.