Effective target aware visual navigation for UAVs

Abstract

In this paper we propose an effective visionbased navigation method that allows a multirotor vehicle to simultaneously reach a desired goal pose in the environment while constantly facing a target object or landmark. Standard techniques such as Position-Based Visual Servoing (PBVS) and Image-Based Visual Servoing (IBVS) in some cases (e.g., while the multirotor is performing fast maneuvers) do not allow to constantly maintain the line of sight with a target of interest. Instead, we compute the optimal trajectory by solving a non-linear optimization problem that minimizes the target reprojection error while meeting the UAV's dynamic constraints. The desired trajectory is then tracked by means of a real-time Non-linear Model Predictive Controller (NMPC): this implicitly allows the multirotor to satisfy both the required constraints. We successfully evaluate the proposed approach in many real and simulated experiments, making an exhaustive comparison with a standard approach.

Non-linear model predictive control with adaptive time-mesh refinement

“…With a reduced computational effort, the solver can run at 10 Hz, thus limiting the vehicle's capacities to safely navigate in dynamic environments. These issues are confirmed in our previous work [19], where two different NMPCs have been adopted in order to handle an Optimal Visual Servoing problem. The first serves for estimating the optimal trajectory in terms of a target re-projection error, while the second is employed as a high-level controller.…”

“…Recently, MPC approaches have been employed in several robotics contexts [10][3][1][14] [16], thanks to on-board increased computational capabilities. In [12] and [20] ACADO, a framework for fast Nonlinear Model Predictive Control (NMPC), is presented; [13] and [19] use ACADO for fast attitude control of an Unmanned Aerial Vehicle (UAV). Such a framework has been further improved in [25] by adding a code generator for embedded implementation of a linear MPC, based on an interior-point solver.…”

“…UAV frames represent trajectory points along the prediction horizon T : the blue frames represent the refined portion of the lattice, while the red ones correspond to the coarser rest of trajectory. does not allow real-time computation, thus making NMPC impractical in real-world applications 1 [19] [21].…”

PAMPC: Perception-Aware Model Predictive Control for Quadrotors

“…In [18], the authors proposed a two-step approach for target-aware visual navigation. First, position-based visual servoing was exploited to find a trajectory minimizing the reprojection error of a landmark of interest.…”

Joint Vision-Based Navigation, Control and Obstacle Avoidance for UAVs in Dynamic Environments

“…A vision-based UAV needs three main components to navigate effectively in an environment populated of possibly dynamic obstacles: 1) A reactive control strategy, to accurately track a desired trajectory while reducing the motors effort; 2) A reliable collision avoidance module, to safely navigate the environment even in presence of dynamic, unmodeled obstacles; 3) An adaptive, perception aware, on-line planner, to support the vision-based state estimation or to constantly keep the line-of-sight with a possible reference target. A wide literature is available addressing individually or in pairs these tasks (among others, [21], [11], [25]). However, they have rarely been addressed together, particularly when dealing with unexpected and moving obstacles.…”