Model Predictive Control for Aerial Collision Avoidance in Dynamic Environments

Abstract

Autonomous navigation in unknown environments populated by humans and other robots is one of the main challenges when working with mobile robots. In this paper, we present a new approach to dynamic collision avoidance for multi-rotor unmanned aerial vehicles (UAVs). A new nonlinear model predictive control (NMPC) approach is proposed to safely navigate in a workspace populated by static and/or moving obstacles. The uniqueness of our approach lies in its ability to anticipate the dynamics of multiple obstacles, avoiding them in real-time. Exploiting active set algorithms, only the obstacles that affect to the UAV during the prediction horizon are considered at each sample time. We also improve the fluency of avoidance maneuvers by reformulating the obstacles as orientable ellipsoids, being less prone to local minima and allowing the definition of a preferred avoidance direction. Finally, we present two real-time implementations based on simulation. The former demonstrates that our approach outperforms its analog static formulation in terms of safety and efficiency. The latter shows its capability to avoid multiple dynamic obstacles.

A Real-Time Approach for Chance-Constrained Motion Planning With Dynamic Obstacles

“…There are two main strategies in the literature to encapsulate an obstacle's space: Using a convex polyhedron [4], [10] (e.g. a cuboid), or a single differentiable surface [5], [6], [15] (e.g. an ellipsoid).…”

“…Alternatively, an obstacle can be bounded by a single differentiable surface (sphere, cylinder, ellipsoid, etc.) to be included as a nonlinear constraint of the optimization problem [15]. This results in a comparatively low-dimension nonlinear program (NLP), which can be solved efficiently by gradient-based solvers [12].…”

“…This results in a comparatively low-dimension nonlinear program (NLP), which can be solved efficiently by gradient-based solvers [12]. Even though this solution cannot guarantee global optimality, its reduced computational cost makes this strategy to be widely adopted in most time-critical motion planning tasks, such as model predictive control for aerial robots [5], [6], [15].…”

Deep Reinforcement Learning for End-to-End Local Motion Planning of Autonomous Aerial Robots in Unknown Outdoor Environments: Real-Time Flight Experiments

“…This increases the autonomy of the robotic systems and makes them avoid dynamic obstacles. For instance, in [ 13 ], the authors proposed an approach based on nonlinear model predictive control (NMPC) for dynamic collision avoidance for a multi-rotor unmanned aerial vehicle. The presented technique combines the optimal path planning and optimal control design into a unified optimization problem; it was tested only in simulation, and it does not consider the model uncertainty and external disturbances.…”

Perception-constrained and Motor-level Nonlinear MPC for both Underactuated and Tilted-propeller UAVS

“…Recently, MPC has gained popularity in aerial robotics, and some work started to incorporate perception-based constraints [2]- [5] directly into the multi-rotor control architecture, since these constraints can be expressed and implemented as a general inequality, like any other constraint acting on the system. In particular, MPC is a model-based and optimization-based control technique using the dynamic model of the system to predict its behavior over a finite receding horizon.…”