---

In this paper, the problem of path planning for a special hyper-redundant manipulator with lockable joints is solved using particle swarm optimization. Two strategies have been used. In the first strategy, the path-planning problem is solved in two stages; first, the inverse kinematics of manipulator is solved with continuous PSO and then the sequence of switching is optimized by modified discrete PSO. The results of discrete PSO are compared to discrete GA to show the effectiveness of the discrete PSO. In order to implement multi-objective optimization in inverse kinematics calculation, linear fitness function and Vector Evaluated PSO (VEPSO) are used. In the second strategy, a novel approach in particle swarm optimization is proposed. In this approach, the integer and continuous value numbers are joined together in a single particle to form a hybrid particle. The results of two strategies were compared to show the convergence speed and performance of the second strategy which was proposed here.

---

In this paper, the kinematic path planning of a special hyper-redundant manipulator with lockable joints is studied. In this manipulator the extra cables are replaced by a locking system to reduce weight of the structure and the number of actuators. In this research, the particle swarm optimization is used for path planning. In addition, the kinematic constraints such as joint limits are considered. In the first part of this paper, the minimum switch path planning is solved. This kind of path planning will decrease the vibration and energy consumption and increase the accuracy of manipulator. To validate the result, an innovative test is designed. According to the test results, the performance of the proposed method is shown. In the second part of the paper, the minimum time trajectory planning is studied based on the bang-bang theory. The inverse kinematic of manipulator is calculated such that the sum of legs length changes is decreased. Finally, the result of trajectory planning obtained from particle swarm optimization are compared to simulated annealing optimization results to confirm the performance and correctness of results.

---

In this paper a method for online identification of satellite moment of inertia tensor parameters based on recursive least squares method, is presented. It is assumed that the satellite actuators are three orthogonal reaction wheels. Dynamic equations of the satellite are extracted in a special manner. The only available sensor is a three axes rate gyro which measures the angular velocity of satellite in the body coordinate system. Due to existence of noise in this sensor, the regressor matrix used in least squares method, changes stochastically. So in this case, the classic least squares method is not useful, and it cannot converge. For solving this problem, a modified least squares method with robust scheme is presented and its stability is proved using Lyapunov stability theory. The presented method can be used online in presence of measurement noise and other sensor imperfections. Simulation results have shown that this method can identify inertia parameters of the satellite with less than 3 percent error comparing to real parameters before and after changes.

---

Exoskeleton is a machine composed of a wearable anthropomorphic structure which noticeably magnifies user's might via its actuators. In this research, dynamic modeling and control system design for a lower limb type of this robot were done. In the literature at most a 1 DOF part of the robot is modeled and controlled which doesn't give a good insight on how all of the robot parts are controlled simultaneously. First, a suitable structure was chosen similar to that of UC Berkeley's BLEEX project. Then dynamic equations were derived in sagittal plane using the Newton-Euler method. By an experiment using Xsens system, gate kinematics data were measured and the inverse dynamics was simulated both in SimMechanics and on the model in MATLAB that proved accuracy of the derived model. Impedance control was investigated and some corrective remarks were included in that algorithm. Using this method the robot was controlled. It stabilized the system and the robot followed user's movement exactly. While a load of 50 kilograms was carried, mostly moments of less than 1 (Nm) were applied at each interface among man and robot.

---

Exoskeleton robot has attracted attention of many researchers because it realizes an old aspiration to attain a machine which is worn by man and maximizes his might via its powerful actuators. Thorough coordination between robot movements and that of user constitutes the greatest complexity of exoskeleton technology. Investigations showed that impedance control (IC) is suitable for this application but the present forms of IC are mostly dedicated to industrial robots which have significant differences with exoskeleton. In this article a versatile form of IC for the mentioned application is developed. Besides, according to perpetual uncertainties in load and measured force signals, for the first time an adaptive method for IC of exoskeleton robot is implemented. Simulating operation of robot in tracking user's walking motion while carrying a load of 50 (Kg) and with uncertainties in load and measured forces, proves efficiency of the proposed control method. Tracking error during simulation is almost zero and torques needed at interfaces are immaterial.

---

Motion control of a planar nonholonomic system with four DoF is addressed in this paper. Three actuators are responsible for shape control of this system. Furthermore, assuming no external forces and zero angular momentum, imposes a nonholonomic constraint to the problem. First it is shown that although the simplified equations of motion for this system, could be converted to Heisenberg and chained-form systems, the conventional control methods for these systems, may not be applied to the considered problem. Then, using sliding modes and online path planning, two different closed-loop control laws are designed for bringing the system to and stabilizing around any desired equilibrium state started from any initial condition. Simulation results, show the efficiency of the proposed methods.

---

In this paper the goal is to identify the parameters of the Lorenz chaotic system, based on synchronization of identical systems using fractional calculus. The method which is used for synchronization is come from Lyapunov stability theorem and then by using fractional dynamics, control laws are improved. To this end, a Lyapunov function is presented and based on the Lyapunov stability theory and asymptotic stability criteria, some adaptation laws to estimate unknown parameters of the system are proposed. The introduced method is applied to the Lorenz chaotic system and since the goal is identification, all the parameters of the system are taken unknown. Using invariant set theory, it is proved that the parameter estimation errors converge to zero. Then the results of numerical simulations are shown and discussed and it is observed that fractional calculus has an essential effect on reducing fluctuations and settling time of the parameters convergence. At the end, the stability of the system by using fractional adaptation law is discussed. It is shown that the asymptotic stability of the synchronization error dynamics is proved using the fractional adaptation law, and this is confirmed through simulation.

---

In this article, stability analysis for Stochastic Piecewise Affine Systems which are a subclass of stochastic hybrid systems is investigated. Here, additive noise signals are considered that does not vanish at equilibrium points. These noises will prohibit the exponential stochastic stability discussed widely in literature. Also, the jumps between the subsystems in this class of stochastic hybrid systems are state-dependent which make stability analysis more complex. The presented theorem considering both additive noise and state-dependent jumps, gives upper bounds for the second stochastic moment or variance of Stochastic Piecewise Nonlinear Systems trajectories and guarantees that stable systems have a steady state probability density function. Then, linear case of such systems is studied where the stability criterion is obtained in terms of Linear Matrix Inequality (LMI) and an upper bound on state covariance is obtained for them. Next, to validate the proposed theorem, solving the Fokker Plank equations which describes the evolution of probability density function, is addressed. A solution for the problem of boundary conditions that arises from jumps in this class of systems is given and then with finite volume method the corresponding partial differential equations are solved for a case study to check the results of the presented theorem numerically.

---

Inertial navigation system has drift error in underwater applications. Use of DVL with Kalman filter for position and attitude correction is common. Using velocity data decreases drift error in position estimation but this error exists and increases linearity with time. In this article the navigation system consists of inertial measurement unit (IMU) and a Doppler velocity log (DVL) along with depth sensor. With use of magnetic field measurement and earth magnetic field map a new measurement is generated. Discrete extended Kalman filter with indirect feedback is used for tightly coupled integrated navigation algorithm. This algorithm is based on inertial navigation error dynamics. This paper demonstrates the effectiveness of algorithm through simulation. The procedure of simulation is done by sensor data generation. Arbitrary trajectory with specific kinematic characteristic (linear and angular velocity and acceleration) is generated. Sensor data by adding noise and bias to kinematic characteristic of trajectory is produced. Simulation results reveal that the new algorithm with use of magnetic data and earth magnetic field map decreases the drift error with comparison to conventional INS-DVL integrated navigation algorithm.

---

In this paper, we investigate chaos in attitude dynamics of a rigid satellite in an elliptic orbit analytically and numerically. The goal in the analytical part is to prove the existence of chaos and then to find a relation for the width of chaotic layers based on the parameters of the system. The numerical part is aimed at validating the analytical method using the Poincare maps and the plots obtained on the sensitivity to initial conditions. For this end, first, the Hamiltonian for the unperturbed system is derived. This Hamiltonian has three degrees of freedom due to the three-axis free rotation of the satellite. However, the unperturbed attitude dynamics has two first-integrals of motion, namely, the energy and the angular momentum. Next, we use the Serret-Andoyer transformation and reduce the unperturbed system Hamiltonian to one-degree of freedom. Then, the gravity gradient perturbation due to moving in an elliptic orbit is approximated in Serret-Andoyer variables and time. Due to this approximation and simplification, the system Hamiltonian transforms to a one-degree-of-freedom non-autonomous one. After that, Melnikov’s method is used to prove the existence of chaos around the heteroclinic orbits of the system. Finally, a relation for calculating the width of chaotic layers around the heteroclinic orbits in the Poincare map of the Serret-Andoyer variables is analytically derived. Results show that the analytical method gives a good approximation of the width of chaotic layers. Moreover, the results show that the analytical method is accurate even for orbits with large eccentricities.

---

According to numerous capabilities and increasingly military and commercial applications of radio controlled helicopters, many investigations are being performed on these unmanned aircraft vehicles. Due to nonlinear, complex, unstable and coupled dynamic system and also existing limitations on manual control, the ability of automatic control of these vehicles has gained great importance. In this paper, in addition to investigating different methods of unmanned helicopters dynamic modeling, a multi-level simulator environment has been designed and implemented for flight performance analysis and effects of different parameters have been investigated. The main importance and innovation of present simulator is in possibility of dynamic flight simulation of helicopter using different theories for applications like control system design, performance analysis and real flight simulation. The main difference of the utilized methods is in theories and assumptions applied in main rotor and its flapping dynamics modeling. For each level, Kalman filter and control system design have been performed and preliminary results show the acceptable performance of estimator and controller systems. Considering the complexity of real unmanned helicopter behavior compared to previously performed models, the proposed multi-level simulator can be used as an appropriate tool for the first step before real flight tests.

---

In this study, design of a novel robust control method for three-link underactuated biped robot which can satisfy appropriate constrains on the robot and cause the stability and rhythmic movement of the robot, is presented. Due to the wide use of sensors and actuators in mechanisms and robots, and existence of noise and also uncertainty in the system components or other error stemming from unmolded dynamics in the system or unwanted disturbances acting on them, there is an essential need for employment of robust control methods. In order to apply locomotion’s constrains to the system, the feedback linearization method is used. Additionally this method is combined with the sliding mode method to obtain a robust control method. The other purpose of the study is the complete elimination of chattering phenomenon. To this end, the control method is combined with the backstepping method. Finally, the exponential stability of the method, which is called FLBS, is proved, and the stability of the obtained walking cycle is shown using the Poincare map. In the robot modeling procedure, the contact between the swing leg and the ground is considered to be rigid and instantaneous. The underactuated nature of the robot and the importance of the role of contact in stabilizing the robot movement are taken into account in this study. Finally, simulations based on this control method are performed which show the exponential stability of robot movement, elimination of chattering and robustness against either disturbances or uncertainties.

---

Human walking is one of the most robust and adaptive locomotion mechanisms in nature, involves sophisticated interactions between neural and biomechanical levels. It has been suggested that the coordination of this process is done in a hierarchy of levels. The lower layer contains autonomous interactions between muscles and spinal cord and the higher layer (e.g. the brain cortex) interferes when needed. Inspiringly, in this study, we present a hierarchical control architecture in order to control under-actuated and high degree of freedom systems with limit cycle behavior and it is implemented for the walking control of a 3-link biped robot. In this architecture, the system is controlled by independent control units for each joint at the lower layer. In order to stabilize the system, these units are driven by a sensory feedback from the posture of the robot. A central stabilizing controller at the upper layer arises in case of failing the units to stabilize the system and take the responsibility of training the lower layer controllers. We show that using this architecture, a highly unstable system can be stabilized with identical simple controller units even though they do not have any feedback from all other units and the robot.

---

Nowadays, nano-precision positioning stages, have a special position and are used in a variety of applications, such as taking pictures and taking particles of the surface. In this paper,some observers for a nano-precision positioning platform are designed based on three different types of neural networks. The simulated platform was designed at Sharif University of Technology and, based on the system's final requirement for the feedback signal for use in the control rule, neural network observers were designed. In previous studies, the comsol model of the positioning system has been obtained. At this step, the neural network has used the Comsol model and the system has been trained for a sum of a number of sinusoidal functions, and its generalizability has been investigated for ramp input. Neural networks used include, respectively, a multi-layer perceptron network, a radial basis function network and a support vector regression network. By performing simulations, it has been seen that the multi-layer perceptron network and the radial basis function network yielded a good response with low error, but the support vector regression network has a relatively high error.

---

In this study, a dynamic based control algorithm for a six-link quadruped locomotion is proposed. Up to now, a lot of robotic scientists have researched in quadruped locomotion but most of their researches are based on modeling of robot and its surrounding. Such methods are not able to generate a stable locomotion when the surrounding changes a little. So this is important to propose a dynamic based control algorithm. The algorithms that can guarantee the stability are classified to two categories of dynamic based and trajectory based methods. The trajectory based algorithms need detailed information of gait and surrounding which is not necessarily available. But the dynamic based algorithms use some geometric constraints to reach a stable controller. These geometric constraints generate the proper gaits. So in this study by employing the dynamic based control algorithm, we proposed a controller for generating the Trot and Pace gait on a straight and flat path for quadruped robot locomotion. Given that the quadruped robot has four degrees of freedom so three geometric constraints are needed to provide a rhythmic locomotion. In this study we showed that for step generating by quadruped robot, both the appropriate initial conditions for angular velocities and presence of a point mass on the neck of the robot are needed. Also in this study the stability of quadruped locomotion has been proved using Poincaré return map.