Results

This section is dedicated to the presentation of simulation and experimental results obtained in the framework of the MAMBO project.
In addition, we have also chosen to present experimental results showing the expertise and skills acquired by the two research teams ARMEN and RAINBOW in the field of UAV control (alone or rigidly linked).

Results of the MAMBO project

  • This video shows the first experimental results obtained with a robot manipulator manipulating a flexible tether and whose control is performed by visual servoing minimizing the difference between the actual characteristics and the desirable characteristics (desired configuration of the tether to be achieved). These first experimental results show the effectiveness of a new visual servoing approach developed in the framework of the MAMBO project and resulting from the thesis of Lev Smolentsev (RAINBOW team).
    The proposed visual servoing approach controls the deformation of a flexible tether from visual data provided by an RGB-D camera. The string is made of an elastic material with negligible torsional and shear stresses. Its shape is modeled by a second order polynomial curve as well as the orientation of the plane containing the curve. The visual characteristics considered in the visual servoing are the polynomial coefficients and the orientation angle of the plane. We derived the analytical expression of the interaction matrix that relates the variation of the visual features to the velocities of the attachment points. An image processing algorithm was also developed to extract the current features fitting the polynomial to the cable in real time from the observed point cloud.
  • Clamped-released (Video 1):
    In this video, we exemplify the reduced dynamics of the rod model used in the MAMBO project. Its dynamics are reduced as the rod is not allowed to move with its proximal end as it is anchored to the ground (in the video at the top). In its undeformed configuration, the beam is straight and stress-free. After applying a deflecting force at the rod’s tip (in the video at the bottom), the force is removed and the rod is subject to its internal undamped restoring forces. Thus, a harmonic oscillation occurs with the rod tip moving on an elliptical path around the origin.
    This model helps us verify both the response of the rod under a static tip force as well as the dynamic oscillating behavior.
  • Free-free (Video 2):
    This scenario depicts a fully free-free rod i.e., neither its proximal (on the right; red path) nor its distal end (on the left; orange path) are geometrically constrained / fixed yet only subjected to external wrenches. These wrenches are composed of forces and moments in such a way that the rod is subject to a twisting moment i.e., a moment about its neutral axis (depicted in red), as well as deflecting moments and forces causing a circular path of the distal end (in orange) around the proximal end (in red).
    We use this scenario to evaluate the model’s correctness and performance for the later use case of a doubly manipulated rod.
  • Free-Free (Video 3):

This scenario shows the beam flying freely through space with the same set of wrenches as applied in the case of video 2. This time, however, the rod’s material properties have been changed to reflect a less flexible beam i.e., increased elongation and bending stiffness. The resulting behavior is a spatial motion similar to a catenary and increasing velocities.

We further used this scenario as a test-bench for comparing various explicit and implicit time integration schemes in order to improve numerical execution times. Using an implicit time integration scheme, we can reduce simulation times drastically down to ~20% of the integration time with an explicit integration scheme.

Previous experimental results (Before the MAMBO project was launched)

ARMEN Team

  • Cable-Suspended Parallel Robots Actuated by Quadrotors UAVs

Cable-Suspended Parallel Robots Actuated by Quadrotors UAVs [Erskine, J., Chriette, A., and Caro, S. (February 22, 2019). “Wrench Analysis of Cable-Suspended Parallel Robots Actuated by Quadrotor Unmanned Aerial Vehicles.” ASME. J. Mechanisms Robotics. April 2019; 11(2): 020909. https://doi.org/10.1115/1.4042513]

  • Flying parallel robots

Damien Six, Sébastien Briot, Abdelhamid Chriette, Philippe Martinet. Dynamic modelling and control of flying parallel robots. Control Engineering Practice, Volume 117, 2021, 104953, ISSN 0967-0661.

RAINBOW team

  • B. Penin, P. Robuffo Giordano, F. Chaumette.Vision-Based Reactive Planning for Aggressive Target Tracking while Avoiding Collisions and Occlusions. IEEE Robotics and Automation Letters 2018 (Best 2018 RAL Paper Award).

In this work we studied the online generaton of time-optimal trajectories for tracking a moving target by a quadrotor with an onboard camera. We considered actuation limits and realistic sensing constraints (camera limited fov and possible occlusions). The proposed approach continuously replans optimal trajectories during flight which provides robustness against model and sensing uncertainties.

  • B. Penin, P. Robuffo Giordano, F. Chaumette. Minimum-Time Trajectory Generation Under Intermittent Measurements. IEEE Robotics and Automation Letters 4(1):153-160, January 2019.

In this work we proposed a minimum-time planning strategy for a quadrotor UAV towards a goal state under the assumption that the quadrotor needs to estimate its state from landmark measurements that may not always be available (e.g., because too far away or occluded). The planning strategy takes into account the evolution of the uncertainty in estimating the UAV state, and finds a feasible trajectory from start to end that maintains the uncertainty below a desired threshold despite the possible temporary lack of measurements during motion.

  • Image-based visual servoing of a Mikrokopter drone.

Comments are closed.