Publications

A control system based on multiple sensors is proposed for the safe collaboration of a robot with a human. New constrained and contactless human-robot coordinated motion tasks are defined to control the robot end-effector so as to maintain a desired relative position to the human head while pointing at it. Simultaneously, the robot avoids any collision with the operator and with nearby static or dynamic obstacles, based on distance computations performed in the depth space of a RGB-D sensor. The various tasks are organized with priorities and executed under hard joint bounds using the Saturation in the Null Space (SNS) algorithm. A direct human-robot communication is integrated within a mixed reality interface using a stereo camera and an augmented reality system. The proposed system is significant for on-line, collaborative quality assessment phases in a manufacturing process. Various experimental validation scenarios using a 7-dof KUKA LWR4 robot are presented.

We propose a new approach for executing the main Cartesian tasks assigned to a redundant robot while guaranteeing whole-body collision avoidance. The robot degrees of freedom are fully utilized by introducing relaxed constraints in the definition of operational and collision avoidance tasks. Desired priorities for each task are assigned using the so-called Task Priority Matrix (TPM) method, which is independent from the redundancy resolution law and handles efficiently switching of priorities. To ensure smooth motion during such task reordering, a control scheme with a suitable task allocation algorithm is developed at the acceleration level. The proposed approach is validated with MATLAB simulations and an experimental evaluation using the 7-dof KUKA LWR manipulator.

We consider the known phenomenon of torque oscillations and motion instabilities that occur in redundant robots during the execution of sufficiently long Cartesian trajectories when the joint torque is instantaneously minimized. In the framework of on-line local redundancy resolution methods, we propose basic variations of the minimum torque scheme to address this issue. Either the joint torque norm is minimized over two successive discrete-time samples using a short preview window, or we minimize the norm of the difference with respect to a desired momentum-damping joint torque, or the two schemes are combined together. The resulting local control methods are all formulated as well-posed linear-quadratic problems, and their closed-form solutions generate also low joint velocities while addressing the primary torque optimization objectives. Stable and consistent behaviors are obtained along short or long Cartesian position trajectories, as illustrated with simulations on a 3R planar arm and with experiments on a 7R KUKA LWR robot.

The problem of minimizing the transfer time along a given Cartesian path for redundant robots can be approached in two steps, by separating the generation of a joint path associated to the Cartesian path from the exact minimization of motion time under kinematic/dynamic bounds along the obtained parametrized joint path. In this framework, multiple sub-optimal solutions can be found, depending on how redundancy is locally resolved in the joint space within the first step. We propose a solution method that works at the acceleration level, by using weighted pseudoinversion, optimizing an inertia-related criterion, and including null-space damping. Several numerical results obtained on different robot systems demonstrate consistently good behaviors and definitely faster motion times in comparison with related methods proposed in the literature. The motion time obtained with our method is reasonably close to the global time-optimal solution along same Cartesian path. Experimental results on a KUKA LWR IV are also reported, showing the tracking control performance on the executed motions.

In the framework of Human-Robot Collaboration, a robot and a human operator may need to move in close coordination within the same workspace. A contactless coordinated motion can be achieved using vision, mounting a camera either on the robot end-effector or on the human. We consider here one instance of such a visual coordination task, with the robot end-effector that should maintain a prescribed position with respect to a moving RGB-D camera while pointing at it. For the 3D localization of the moving camera, we compare three different techniques and introduce some improvements to the best solution found for our application. For the motion tracking problem, we introduce a relaxed version of the pointing part of the task. This allows to take advantage of the redundancy of the robot, distributing the control effort over the available degrees of freedom. The effectiveness of the proposed approach is shown by V-REP simulations and experiments with the 7-dof KUKA LWR manipulator.

Optimal control for kinematically redundant robots is addressed for two different optimization problems. In the first optimization problem, we consider the minimization of the transfer time along a given Cartesian path for a redundant robot. This problem can be solved in two steps, by separating the generation of a joint path associated to the Cartesian path from the exact minimization of motion time under kinematic/dynamic bounds along the obtained parametrized joint path. In this thesis, multiple sub-optimal solutions can be found, depending on how redundancy is locally resolved in the joint space within the first step. A solution method that works at the acceleration level is proposed, by using weighted pseudoinversion, optimizing an inertia-related criterion, and including null-space damping. The obtained results demonstrate consistently good behaviors and definitely faster motion times in comparison with related methods proposed in the literature. The motion time obtained with the proposed method is close to the global time-optimal solution along the same Cartesian path. Furthermore, a reasonable tracking control performance is obtained on the experimental executed motions. In the second optimization problem, we consider the known phenomenon of torque oscillations and motion instabilities that occur in redundant robots during the execution of sufficiently long Cartesian trajectories when the joint torque is instantaneously minimized. In the framework of on-line local redundancy resolution methods, we propose basic variations of the minimum torque scheme to address this issue. Either the joint torque norm is minimized over two successive discrete-time samples using a short preview window, or we minimize the norm of the difference with respect to a desired momentum-damping joint torque, or the two schemes are combined together. The resulting local control methods are all formulated as well-posed linear-quadratic problems, and their closed-form solutions generate also low joint velocities while addressing the primary torque optimization objectives. Stable and consistent behaviors are obtained along short or long Cartesian position trajectories. For the two addressed optimization problems in this the- sis, the results are obtained using three different robot systems, namely a 3R planar arm, a 6R Universal Robots UR10, and a 7R KUKA LWR robot.