Abstract:

With regarding to the dynamic obstacle avoidance problem in the dynamic environment encountered by the current unmanned aerial vehicles (UAVs), the kinetics of both the UAV and dynamic obstacles are established on the basis of the appropriate hypothesis, and the terminal constraints, control limitations as well as safe avoidance are taken into consideration. With the minimal energy consumption as the performance index, the dynamic obstacle avoidance problem is described mathematically. Then, with regarding to the terminal constraints and control limitations, an initial trajectory is generated according to the Optimized Model Predictive Static Programming (OMPSP). Against to the inequality constraint resulting from the dynamic obstacle avoidance, the slack variables are introduced and combined with the sliding mode control, and their dynamics are designed to implement the avoidance trajectory for single or multiple dynamic obstacles simultaneously. Eventually, the trajectory is further optimized by Receding Horizontal Differential Dynamic Programming (RHDDP). Consequently, a near optimal trajectory which satisfies multiple constraints and is capable of avoiding dynamic obstacles is developed.

Key words:  Unmanned aerial vehicle, Dynamic obstacle avoidance, Optimized model predictive static programming, Slack variables, Near optimal trajectory