We present the design, motion planning and control of an aerial manipulator for non-trivial physical interaction tasks, such as pushing while sliding on curved surfaces-a task which is motivated by the increasing interest in autonomous non-destructive tests for industrial plants. The proposed aerial manipulator consists of a multidirectional-thrust aerial vehicle-to enhance physical interaction capabilities-endowed with a 2-DoFs lightweight arm-to enlarge its workspace. This combination makes it a truly-redundant manipulator going beyond standard aerial manipulators based on collinear multirotor platforms. The controller is based on a PID method with a 'displaced' positional part that ensures asymptotic stability despite the arm elasticity. A kinodynamic task-constrained and control-aware global motion planner is used. Experiments show that the proposed aerial manipulator system, equipped with an Eddy Current probe, is able to scan a metallic pipe sliding the sensor over its surface and preserving the contact. From the measures, a weld on the pipe is successfully detected and mapped.
Abstract-This paper presents a new method to address the problem of task-constrained motion planning for aerial manipulators. We propose a control-aware planner based on the paradigm of tight coupling between planning and control. Such paradigm is especially useful in aerial manipulation, where the separation between planning and control is not advisable. The proposed sampling based motion planner uses a controller composed of a second-order inverse kinematics algorithm and a dynamic tracker, as a local planner, thus allowing a more natural consideration of the closed-loop system dynamics. For task constrained motions, this method lets to i) sample directly in the reduced and more relevant task space, ii) predict the behavior of the controller avoiding motions that bring to singularities or large tracking errors, and iii) guarantee the correct execution of the maneuver. The method is tested in simulation for a multidirectional-thrust vehicle endowed with a two-DoF manipulator. The proposed approach is very general, and could be applied to ground and underwater robotic systems to perform manipulation or inspection tasks.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.