In this paper, we present the work related to the application of a visual odometry approach to estimate the location of mobile robots operating in off-road conditions. The visual odometry approach is based on template matching, which deals with estimating the robot displacement through a matching process between two consecutive images. Standard visual odometry has been improved using visual compass method for orientation estimation. For this purpose, two consumer-grade monocular cameras have been employed. One camera is pointing at the ground under the robot, and the other is looking at the surrounding environment. Comparisons with popular localization approaches, through physical experiments in off-road conditions, have shown the satisfactory behavior of the proposed strategy.

As every real mechanical hand has a limited accuracy, the grasp planning process must be prepared to cope with unavoidable positioning errors. The concept of independent contact regions (ICRs) was proposed to deal with this issue by computing for each finger an ICR on the object's boundary such that each finger can be placed anywhere in its ICR to guarantee a force closure grasp. Existing methods for computing ICRs of a polygon requires that each ICR must lie on a single edge of the polygon. This constraint severely limits the size of computed ICRs, especially when the input polygon contains only small edges (e.g., when the polygon is used for representing a curve object). This paper proposes a method for computing the optimal ICRs for frictional two-fingered grasp of a polygon such that each ICR is allowed to extend across consecutive edges of the polygon. The time complexity of the method is O(n2log n), where n is the number of edges of the polygon. Implementation results using several test polygons are presented to exhibit effectiveness of the method.

Matched filtering optimally estimates the arrival time for a sonar sensor by correlating received signals with templates. This paper presents a sonar ring with continuous matched filtering on 48 receiver channels sampled at 500kHz. The design dynamically switches the matched filter templates to account for pulse shape variations with range. To achieve real-time, low-latency and optimal performance, processing is implemented on an field-programmable gate array (FPGA) transmitting sonar pulses (2 periods of a 45kHz sine wave) at repetition rate of 30-Hz to 5.7-m range. The paper describes the removal of secondary peaks of the correlation output of matched filtering and template selection. Results include sonar maps, accuracy measurements and localization of weak targets.

A biomimetic underwater vehicle, which is propelled by two undulating long-fins, is introduced in this paper. The undulating or oscillating movements of symmetrical long-fins cause the complex locomotion of biomimetic underwater vehicle. For convenience, three motion modes are proposed and considered firstly. Then an inertial unit is installed for collection of accelerations and angular velocity. The underwater vehicle's MIMO model is reduced into a SISO model by some simplifications. A sine wave function deduced from the long-fin's time-varying membrane is proposed and used as the input of the biomimetic underwater vehicle ARMA model, and velocity or angular velocity is considered as the model output. The algorithms based on recursive weighted least squares are applied for model parameter identification. Experiments carried out with a long-fin propelled underwater vehicle. The experimental results show that the proposed methods can build valid locomotion models for three motion modes efficiently.

In this paper a 3-degree of freedom compliant parallel positioning stage utilizing flexure hinges is explored for nanoimprint lithography. The performance of the stage is extensively analyzed using a pseudo-rigid body model and finite element method. The position and velocity models are established. Accordingly, the stiffness at driving point of the active arm is obtained on the basis of Castigliano's first theorem. The system stiffness of the compliant stage is explored using the compliant matrix methodology and matrix transformation, and the influence of the geometry parameters on stiffness factors in three directions has been graphically evaluated as well. Finite element analysis has been conducted to verify that the established models faithfully predict device performance.

Wire-actuated parallel manipulators and their failures are studied in this paper taking into consideration their failure modes. A methodology for investigating the effect of wire/actuator failures on the force/moment capability of manipulators is presented, and the criteria for full and partial recovery from these failures are established. The methodology is also applicable for the cases that the minimum norm solution for the vector of wire tensions gives a negative value for tension by treating the corresponding wire as failed. The proposed criteria are also valid for the manipulators that utilize hybrid actuation of wires and joints. Three planar wire-actuated parallel manipulators are used as the case study to illustrate the proposed methodology and criteria.

Contrary to the inverse kinematics, the forward kinematics of parallel manipulators involves solving highly non-linear equations and provides more than one feasible end-effector pose, which are called the assembly modes, for a given set of link lengths or joint angles. Out of the multiple feasible solutions, only one solution can be achieved from a certain initial configuration. Therefore, in this paper, the Kohonen's self-organizing map (SOM) is proposed to learn and classify the multiple solution branches of the forward kinematics and then provide a unique real-time solution among the assembly modes. Each solution of the multiple feasible ones is coded using IF-THEN rules based on the values of the passive joint variables. Due to not only the classification but also the associative memory learning abilities of the SOM, the passive joint variables vector, the end-effector pose vector, and this class code are associated with the active joint variables vector constituting the input vector to the SOM in the offline learning phase. In the online testing phase, only the active joint variables vector and the class code are fed to the SOM to obtain the unique end-effector pose vector. The Jacobian matrix calculated at the SOM output layer is used for further fine tuning this output to obtain an accurate end-effector pose vector. Simulations are conducted for 3-RPR and 3-RRR planar parallel manipulators to evaluate the performance of the proposed method. The results proved high accuracy of the desired unique solution in real-time.

Many industrial robots employ closed-loop actuating elements such as the parallelogram mechanism for increased stiffness. Modeling these manipulators for the purpose of calibration presents a challenge due to complex nonlinear couplings between parameters of the chains. The modeling method presented in this paper involves the integration of the open- and closed-loop elements whose errors can be resolved as linear functions of their parameters. As a result, the model is similar to that of a serial-link robot, which makes it possible to use existing well-defined calibration techniques in the area. Simulation and experimental studies on an industrial robot for verifying the correctness and effectiveness of the proposed model are also described.

The effect on failure rates of the way tele-operators interact with mobile robots is investigated. Human tele-operators attempted to move a robot through progressively more complicated environments with reducing gaps, as quickly as possible. Tele-operators used a joystick and either watched robots, while operating them, or used a computer screen to view scenes remotely. Cameras were either mounted on the robot to view the space ahead of the robot or mounted remotely so that they viewed both the environment and robot. Tele-operators completed tests both with and without sensors. Both an umbilical cable and a radio link were used.

This paper investigates the efficiency and properties of limit cycle walking, running, and skipping of a planar, active, telescopic-legged rimless wheel. First, we develop the robot equations of motion and design an output following control for the telescopic-legs' action. We then numerically show that a stable walking gait can be generated by asymmetrizing the impact posture. Second, we numerically show that a stable running gait can be generated by employing a simple feedback control of the control period, and compare the properties of the generated running gait with those of the walking gait. Furthermore, we find out another underlying gait called skipping that emerges as an extension of the walking gait. Through numerical analysis, we show that the generated skipping gaits are inherently stable and are less efficient than the other two gaits.

This paper presents a new algorithm to find the singularity-free cylindrical workspace of parallel manipulators. Because of the limited workspace of parallel manipulators cluttered with different types of singularities, a simple and robust technique to determine continuous singularity-free zones in the workspace of parallel manipulators is required. Here, the largest singularity-free cylinder within the workspace for any prescribed orientation ranging around a reference orientation angle of moving platform is determined. To this end, Particle Swarm Optimization is utilized to find the closest point on the singularity surface to the axis of the cylinder. By implementing the algorithm on 3-RPR planar parallel manipulator, the results show that this algorithm improves the efficiency and leads to significantly larger singularity-free workspace than those reported earlier.

Endowing robots with the ability of skill learning enables them to be versatile and skillful in performing various tasks. This paper proposes a neuro-fuzzy-based, self-organizing skill-learning framework, which differs from previous work in its capability of decomposing a skill by self-categorizing it into significant stimulus-response units (SRU, a fundamental unit of our skill representation), and self-organizing learned skills into a new skill. The proposed neuro-fuzzy-based, self-organizing skill-learning framework can be realized by skill decomposition and skill synthesis. Skill decomposition aims at representing a skill and acquiring it by SRUs, and is implemented by stages with a five-layer neuro-fuzzy network with supervised learning, resolution control, and reinforcement learning to enable robots to identify a sufficient number of significant SRUs for accomplishing a given task without extraneous actions. Skill synthesis aims at organizing a new skill by sequentially planning learned skills composed of SRUs, and is realized by stages, which establish common SRUs between two similar skills and self-organize a new skill from these common SRUs and additional new SRUs by reinforcement learning. Computer simulations and experiments with a Pioneer 3-DX mobile robot were conducted to validate the self-organizing capability of the proposed skill-learning framework in identifying significant SRUs from task examples and in common SRUs between similar skills and learning new skills from learned skills.

Exact knowledge of the position and proper calibration of robots that move by wheels form an important foundation in mobile robot applications. In this context, a variety of sensory systems and techniques have been developed for accurate positioning of differential drive mobile robots. This paper, first, provides a brief overview of mobile robots positioning techniques and then, presents a new benchmark method capable of calibrating mobile robots with differential drive mechanisms to correct systematic errors. The proposed method is compared with the commonly used University of Michigan Benchmark (UMBmark) odometry method. Two sets of comparisons are conducted on six prototyped robots with differential drives. The first set of tests establishes the workability and accuracy that can be achieved with the new method and compares them with the ones obtained from the UMBmark technique. The second experiment compares the performance of a mobile robot, calibrated with either the UMBmark or the new method, for an unseen path. It is demonstrated that the proposed method of calibration is simple to implement, and leads to accuracy comparable to the UMBmark method. Specifically, while the error corrections in both methods are within ±5% of each other, the proposed method requires single straight line motion for calibration, which is believed to be simpler and less timely to implement than the square path motion required by the UMBmark technique. The method should therefore be considered seriously as a new tool when calibrating differential drive mobile robots.