Cartesian Constrained Stochastic Trajectory Optimization for Motion Planning

This paper presents novel extensions of the Stochastic Optimization Motion Planning (STOMP), which considers cartesian path constraints. It potentially has high usage in many autonomous applications with robotic arms, where preservation or minimization of tool-point rotation is required. The original STOMP algorithm is unable to use the cartesian path constraints in a trajectory generation because it works only in robot joint space. Therefore, the designed solution, described in this paper, extends the most important parts of the algorithm to take into account cartesian constraints. The new sampling noise generator generates trajectory samples in cartesian space, while the new cost function evaluates them and minimizes traversed distance and rotation change of the tool-point in the resulting trajectory. These improvements are verified with simple experiments and the solution is compared with the original STOMP. Results of the experiments show that the implementation satisfies the cartesian constraints requirements.

End effector position calculation with the ANN for tapping machine

Nowadays, manufacturers give importance to the production of machines that allow for faster production, reduce labor costs, and minimize operation errors to meet the increasing demand. The search for such machines leads the manufacturing sector to automation. In this study, an automation-supported tapping machine prototype was manufactured. Kinematic equations were used for determining the location of the end effector in Cartesian space, whereas inverse kinematic equations were used for angular positions in joint space relative to positions in Cartesian space. Based on the results of the kinematic equations, the data obtained in certain positions were taught to the system through ANN. The position values for the angles known through the artificial intelligence algorithm were taught to the system. Then, the position coordinates to be reached by this manipulator, which has four degrees of freedom, for the intermediate position coordinate values through artificial neural networks (ANN) have been obtained. It is expected that the device controlled by artificial intelligence will not be affected by the variables in parameter or force changes requiring high working performance. With the control of the positions through ANN, it has been ensured that the position control of the tapping robot manipulator is predicted based on artificial intelligence techniques depending on the angle values of the limbs, and the robot is prevented from going to a position that is on a different trajectory. Accordingly, the robot arm has been made controllable with ANN techniques. With ANN modelling, the position of the end point to perform the tapping process was estimated with high reliability. For future research, a rough simulation was made to see whether the end point would go to a different position in space.

An Extended DMPs framework for Decoupled Quaternions Learning and generalization

Dynamic Movement Primitives (DMPs) as a robust and efficient framework has been studied widely for robot learning from demonstration. Classical DMPs framework mainly focuses on the movement learning in Cartesian or joint space, and can't properly represent end-effector orientation. In this paper, we present an Extended DMPs framework (EDMPs) both in Cartesian space and Riemannian manifolds for Quaternion-based orientations learning and generalization. Gaussian Mixture Model and Gaussian Mixture Regression are adopted as the initialization phase of EDMPs to handle multi-demonstrations and obtain their mean and covariance. Additionally, some evaluation indicators including reachability and similarity are defined to characterize the learning and generalization abilities of EDMPs. Finally, the quaternion-based orientations are successfully transferred from human to the robot, and a real-world experiment is conducted to verify the effectiveness of the proposed method. The experimental results reveal that the presented approach can learn and generalize multi-space parameters under multi-demonstrations.

ADRC-Backstepping-Compensation Control for A Novel Aerial Manipulator System

This paper proposes a fully-actuated control method for a novel aerial manipulation system (AMS). A customized carbon frame structure supports the servo actuators, on which eight propellers group into pairs located. We present kinematics and dynamics modeling of the AMS based on Craig parameter method and recursive Newton–Euler equation, respectively. Then, an Active disturbance rejection control (ADRC)–Backstepping–Compensation controller is designed to control the exact position and orientation of the manipulator in the Cartesian space. Finally, the performance of the system is demonstrated through simulations and outdoor experiments.

Modelling, Design and Validation of a Parallel Kinematic Robot for Additive Manufacture Applications

The additive manufacture is a fabrication process that has taken huge steps in the last decade, even though the first researches and prototypes are around since almost forty years ago. In this article, a design method for a Parallel Kinematics Robot for Additive Manufacturing Applications is proposed. A numerical model is obtained from the kinematics of the robot for which the design, construction and assembly are planned using recycled materials and equipment. The control of the robot is done using open source software, allowing the planning of trajectories in the Cartesian space on a maximum designed cylindrical workspace of 300mm in diameter by 300mm high. At the end of the work the robot was identified, the kinematic model was validated and considerations for future works were given.