Cladding through laser metal deposition is a promising application of additive manufacturing. On the one hand, industrial robots are increasingly used in cladding because they provide wide wrist reorientation, which enables manufacturing of complex geometries. On the other hand, limitations in robot dynamics may prevent cladding of sharp edges and large objects. To overcome these issues, this paper aims at exploiting the residual degrees of freedom granted by the cladding process for the optimization of the deposition orientation. The proposed method optimizes the robot head orientation along a predefined path while coping with kino-dynamic constraints as well as process constraints. Experimental tests and results are reported and used to validate the approach.

Novel kinematic architectures can be alternatives for designing energy efficient robotic systems. In this work, the impact of kinematic redundancies in the energy consumption of a planar PKM, the 3PRRR manipulator, is experimentally verified. Because of the presence of the kinematic redundancies, the inverse kinematic problem presents infinity solutions. In this way, a redundancy resolution scheme based on the Model Predictive Control technique is proposed and exploited. It can be concluded that the energy consumption of the non-redundant parallel manipulator 3RRR for executing predefined tasks can be considerably reduced by the inclusion of kinematic redundancies.

This paper presents a robust adaptive output feedback tracking controller for the flexible-joint robot manipulators to deal with the unknown upper bounds of parameter uncertainties and external disturbances. With applying the singular perturbation theory and integral manifold concept, the complex nonlinear coupled system of the flexible-joint robot manipulators is divided into a slow subsystem and a fast subsystem. A robust adaptive control scheme based on an improved linear parameterization expression is designed for the slow subsystem, and a saturation function is applied in the robust control term to make the torque output smooth. In the meantime, different from the previous approaches, the second-order derivative term of elastic torque is avoided by using the proposed computed torque method, which simplifies the implementation of the fast control law. Moreover, to carry out the whole control system with only position measurements, an approximate differential filter is involved to generate pseudo velocity signals for links and joint motors. In addition, an explicit but strict stability proof of the control system based on the theory of singularly perturbed systems is presented. Finally, simulation results verify the superior dynamic performance of the proposed controller.

This paper proposes a special non-symmetric topology of a 3PRR planar parallel kinematics mechanism, which naturally avoids singularity within the workspace and can be utilized for hybrid kinematics machine tools. Subsequently, single-objective and multi-objective optimizations are conducted to improve the performance. The workspace area and minimum eigenvalue, as well as the condition number of the homogenized Cartesian stiffness matrix across the workspace, have been chosen as the objectives in the optimization based on their relevance to the machining application. The single-objective optimization is conducted by using a single-objective genetic algorithm and a hybrid algorithm, whereas the multi-objective optimization is conducted by using a multi-objective genetic algorithm, a weighted sum single-objective genetic algorithm, and a weighted sum hybrid algorithm. It is shown that the single-objective optimization gives superior value in the optimized objective, while sacrificing the other objectives, whereas the multi-objective optimization compromises the improvement of all objectives by providing non-dominated values. In terms of the algorithms, it is shown that a hybrid algorithm can either verify or refine the optimal value obtained by a genetic algorithm.

This paper presents a new type of forceps that consist of two microgrippers with the capability of direct force sensing, which enables grasping and manipulating forces at the tip of surgical instrument for minimally invasive robotic surgery. For the prototype design of the forceps, a double E-type vertical elastomer with four strain beams is presented, whose force-sensing principle is expounded. Thus, the forceps with the elastomer can be considered a compliant component, which provides tiny displacements that allow large strain, and the overall diameter is 10 mm. The sizes of the elastomer and forceps are successively determined by analyzing the relationship of several parameters and strain. Then, the linearity analysis of strain beams determines the positions to apply gauges for sensing. The two-dimensional force decoupling models for a single microgripper are proposed based on piecewise analytical polynomials of the strain difference and employed to develop a new three-dimensional force nonlinear decoupling algorithm based on double microgrippers, which realizes single-axial grasping and three-axial pulling forces. Finally, the required force-sensing performance of the proposed method is successfully verified in theory using finite-element simulations.

This paper proposes a new path-planning algorithm which is close to the family of bug algorithms. Path planning is one of the challenging problems in the area of service robotics. In practical applications, traditional methods have some limitations with respect to cost, efficiency, security, flexibility, portability, etc. Our proposed algorithm offers a computationally inexpensive goal-oriented strategy by following a smooth and short trajectory. The paper also presents comparisons with other algorithms. In addition, the paper also presents a test bed which is created to test the algorithm. We have used a two-wheeled differential drive robot for the navigation and only a single camera is used as a feedback sensor. Using an extended Kalman filter, we localize the robot efficiently in the map. Furthermore, we compare the actual path, predicted path and planned path to check the effectiveness of the control system.

The constrained coverage path planning addressed in this paper refers to finding an optimal path traversed by a unmanned aerial vehicle (UAV) to maximize its coverage on a designated area, considering the time limit and the feasibility of the path. The UAV starts from its current position to assess the condition of a new entry to the area. Nevertheless, the UAV needs to comply with the coverage task, simultaneously and therefore, it is likely that the optimal policy would not be the shortest path in such a condition, since a wider area can be covered through a longer path. From the other side, along with a longer path, the UAV may not reach to the target in due time. In addition, the speed of UAV is assumed to be constant and as a result, a feasible path needs to be smooth enough to support this assumption. The problem is modeled as an Epsilon-constraint optimization in which a coverage function has to be maximized, considering the constraints on the length and the smoothness of the path. For this purpose, a new genetic path planning algorithm with adaptive operator selection is proposed to solve such a complicated constrained optimization problem. The proposed approach has been compared to some classical approaches like, a modified version of the Artificial Potential Field and a modified version of Dijkstra's algorithm (a graph-based approach). All the methods are implemented and tested in different scenarios and their performances are evaluated via the simulation results.

In this paper, a walking pattern optimization procedure is implemented to yield the optimal heel-strike and toe-off motions for different goal functions. To this end, first, a full dynamic model of a humanoid robot equipped with active toe joints is developed. This model consists of two parts: multi-body dynamics of the robot which is obtained by Lagrange and Kane methods and power transmission dynamic model which is developed using system identification approach. Then, a gait planning routine is presented and consistent parameters are specified. Several simulations and experimental tests are carried out on SURENA III humanoid robot which is designed and fabricated at the Center of Advanced Systems and Technologies located in the University of Tehran. Afterward, a genetic algorithm optimization is adopted to compute the optimal walking patterns for five different goal functions including energy consumption, stability margin, joint velocity, joint torque and required friction coefficient. Also, several parametric analyses are performed to characterize the effects of heel-strike and toe-off angle and toe link mass and length on these five goal functions. Finally, it is concluded that walking pattern without heel-strike and toe-off motions requires less friction coefficient than the pattern with heel-strike and toe-off motions. Also, heavier toe link lowers tip-over instability and slippage occurrence possibility, but requires more energy consumption and joint torque.