This paper suggests one possible mechanism for the biological flexor reflex by emulating a cat's behavior with its robotic counterpart, making it capable of walking around and clearing obstacles autonomously in various environments. A central pattern generator and a hip-to-knee mapping function are employed to realize basic rhythmic motion for a quadrupedal robot. When an input from a contact sensor on the robot's toe is detected, a patterned motion generated by the flexor reflex emerges depending on the location of the bumping phase, and replaces the ongoing rhythmic motion of that leg, causing it to be raised high enough to clear the obstacle. By restricting this reflex to within one cycle time of the walk and only to the bumping leg, rhythm and stability of motion are ensured. Numerical simulations and experimental implementation on a physical quadrupedal robot demonstrate the effectiveness of the proposed method.

A novel method of Zero-Moment-Point (ZMP) compensation is proposed to improve the stability of locomotion of a biped, which is subjected to disturbances. A compensating torque is injected into the ankle-joint of the foot of the robot to improve stability. The value of the compensating torque is computed from the reading of the force sensors located at the four corners of each foot. The effectiveness of the method is verified on a humanoid robot, MANUS-I. With the compensation technique, the robot successfully rejected disturbances in different forms. It carried an additional weight of 390 gm (17% of body weight) while walking. Also, it walked up a 10° slope and walked down a 3° slope.

This paper studies the mechanical configuration and the periodic gaits of multi-legged locomotion systems based on its kinematic and dynamic models. The purpose is to determine the system performance during walking, and the best set of locomotion variables that minimize a set of optimization indices. In this perspective, two kinematic and four dynamic indices are formulated to quantitatively measure the performance of the walking robot. The kinematic indices consist of the perturbation analysis and the locomobility measure, and the dynamic performance indices of the walking robot locomotion are the mean absolute density of energy, the mean power density dispersion, the density of power lost and the mean force at the body-legs interface. A set of model-based simulation experiments reveals the system configuration and the type of movements that lead to a better performance, for a specific locomotion mode, from the viewpoint of the proposed indices.

Genetic algorithm is used to determine the optimal capture points for the multi agents required to grasp a moving generic prismatic object by arresting it in form closure. Thereafter, the agents approach their respective moving goals using a decentralized projective path planning algorithm. Post arrest, the object is guided along a desired linear path to a desired goal point. Form closure of the object is obtained using the concept of accessibility angle. A convex envelop is formed around the object, and the goal points on the object boundary are mapped onto the envelope. The robots approach the mapped goal points first, and then, converge on the actual object. This ensures that the agents reach the actual goal points almost simultaneously, and do not undergo looping at a local concave region. The object is assumed alive while being captured but is assumed compromised thereafter. Post arrest, robots alter their positions optimally around the object to transport it along a desired direction. Frictionless point contact between the object and a robot is assumed. The shape of the mobile robot is considered cylindrical such that it can only apply force along the outward radial direction. Simulation results are presented that illustrate the effectiveness of the proposed method.

This paper presents analysis and experimental verifications of a new robot manipulator with five degrees of freedom developed for the buffing operation of shoes. First, the forward and inverse kinematics are analyzed. Next, an analytic closed-form solution is rigorously derived for the joint angles corresponding to the position and orientation of the end-effector in Cartesian coordinates. A control system, including input/output interfaces and the related electronic system, is designed for the control of the mechanical structure of the buffing robot. Then, peripheral systems integrated with the conveyer, transfer device, and fixture device are designed for the sequential buffing process of shoes. Also, a graphic user interface (GUI) program including the forward/inverse kinematics, control algorithm, and communication program to interact the robot with the peripheral systems is developed by using visual C++ language. A new flexible toolholder (FTH) is proposed to compensate for the excessive applied force between deburring tools and shoes. Finally, the test results are provided to demonstrate the effectiveness of the proposed scheme.

A good practice to ensure high-positioning accuracy in industrial robots is to use joint error maximum mutual compensation (JEMMC). This paper presents an application of JEMMC for positioning of hexapod robots to improve end-effector positioning accuracy. We developed an algorithm and simulation framework in MatLab to find optimal hexapod configurations with JEMMC. Based on a real hexapod model, simulation results of the proposed approach are presented. Optimal hexapod configurations were found using the local minimum of the infinity norm of hexapod Jacobian inverse. JEMMC usage in hexapod robots can improve hexapod end-effector positioning accuracy by two times and more.

A major technical challenge in controlling modular and reconfigurable robots is associated with the kinematics and dynamic model uncertainties caused by reconfiguration. In parallel, conventional model uncertainties such as uncompensated joint friction still persist. This paper presents a modular distributed control technique for modular and reconfigurable robots that can instantly adapt to robot reconfigurations. Under the proposed control method that is based on joint torque sensing, a modular and reconfigurable robot is stabilized joint by joint, and modules can be added or removed without the need to adjust control parameters of the other modules of the robot. Model uncertainties associated with link and payload masses are compensated using joint torque sensor measurement. The remaining model uncertainties, including uncompensated dynamic coupling and joint friction, are compensated by a decomposition-based robust controller. Simulation results have confirmed the effectiveness of the proposed method.

In this paper, we design a position/force controller for cooperative robots during constrained motion. The proposed scheme is based on the knowledge of the manipulator dynamics and does not require measurements of link velocity or end-effector contact forces. A well-known velocity observer and the design of a force observer are used. The validity of the proposed method is verified by means of experimental results.

This paper addresses the dynamic dexterity of a planar 2-degree of freedom (DOF) parallel manipulator with virtual constraint. Without simplification, the dynamic formulation is derived by using the virtual work principle. The condition number of the inertia matrix of the dynamic equation is presented as a criterion to evaluate the dynamic dexterity of a manipulator. In order to obtain the best isotropic configuration of the dynamic dexterity in the whole workspace, two global performance indices, which consider the mean value and standard deviation of the condition number of the inertia matrix, respectively, are proposed as the objective function. For a given set of geometrical and inertial parameters, the dynamic dexterity of the parallel manipulator is more isotropic in the center than at the boundaries of the workspace.

Deformable models have been studied in image analysis over the last decade and used for recognition of flexible or rigid templates under diverse viewing conditions. This article addresses the question of how to define a deformable model for a real-time color vision system for mobile robot navigation. Instead of receiving the detailed model definition from the user, the algorithm extracts and learns the information from each object automatically. How well a model represents the template that exists in the image is measured by an energy function. Its minimum corresponds to the model that best fits with the image and it is found by a genetic algorithm that handles the model deformation. At a later stage, if there is symbolic information inside the object, it is extracted and interpreted using a neural network. The resulting perception module has been integrated successfully in a complex navigation system. Various experimental results in real environments are presented in this article, showing the effectiveness and capacity of the system.

This is the first publication presenting the minihumanoid robot THBIP-2, the second-generation biped of Tsinghua University. It is 70 cm in height and 18 kg in weight with 24 degrees of freedom. This paper mainly addresses its mechatronics system realization, including the conceptual design, actuation system, sensing system, and control system. In addition, a walking stability controller based on zero moment point criterion and the walking simulation are presented. Finally, experiments validate and confirm the efficiency of the design.

This paper is to investigate inherent oscillations problems of potential field methods (PFMs) for nonholonomic robots in dynamic environments. In prior work, we proposed a modification of Newton's method to eliminate oscillations for omnidirectional robots in static environment. In this paper, we develop control laws for nonholonomic robots in dynamic environment using modifications of Newton's method. We have validated this technique in a multi-robot search-and-forage task. We found that the use of the modifications of Newton's method, which applies anywhere C2 continuous navigation functions are defined, can greatly reduce oscillations and speed up the robot's movement, when compared to the standard gradient approaches.