A new working robot has been designed and built on the basis of the concept“Limb Mechanism”that integrates functions of legged locomotion and arm manipulation. As one of feasible structures of the limb mechanism a six-limbed mechanism will be analyzed and evaluated in the aspects of omni-directional mobility. To evaluate the omni-directional mobility, a stroke and a stability margin are examined in all walking directions during six-legged and four-legged locomotion. The paper introduces two types of structures: radial and parallel arrangement of legs, and compares their stroke and stability in all directions during six-legged locomotion. The radial arrangement model will be proved to have the omni-directional mobility even in four-legged locomotion. Based on this evaluation the proto-type robot is designed and built. The developed robot has radially arranged six limbs that can be used for both locomotion and manipulation.
This study focuses on emergence of biped robot locomotion based on the concept called‘Global Entrainment.’Originally, this concept was proposed as a principle of human locomotion emergence; locomotion is self-organized as limit cycles of the coupled dynamical systems, which consist of the neurons, body and environment. In this paper, the author designs the neural system using computer simulations of the biped robot at first. With the designed neural system, the simulated robot steps with oscillatory motion in the lateral plane, and then starts to walk. It is also investigated that the realized locomotion has adaptability to changes of the robot and environment models. The last experiment shows that the real robot is able to walk in the same way as the simulations.
In this decade, a number of various robot contests represented by “RoboCup” have been held in various parts of the world. In some of these contests, such as RoboCup middle-size robot league, fully automated mobile robots are used to compete for their mobility and autonomy. These robots have to plan optimum moving strategy automatically without any human intervention after the game starts. One of the fundamental and crucial faculties for these autonomous robots is the ability of accurate positioning of enemy and themselves. In this paper, a robust positioning system is proposed for soccer robots equipped with an omni-directional camera and wheels. The robot identifies its own position by observing some landmarks placed around the field through an omni-directional camera, and by counting rotation of wheels. To achieve the robust positioning, the proposed method utilizes (1) LMedS method for making correct correspondence between observed directions and landmarks, (2) elimination of miss-correspondence of landmarks using a median filter, (3) position calculation considering the properties of measurement error distributions using nonlinear maximum likelihood method, and (4) data fusion of positions obtained by observing landmarks and dead reckoning using kalman filter.
This paper presents a compliance control method for force sensorless manipulators with harmonic drive gears. The purpose of this study is to realize a compliant motion control in a specified direction and a position control in the other direction simultaneously when the endpoint of the manipulator contacts with an unknown object for quasistatic motions. The proposed method consists of a limitation algorithm to restrict endpoint force to a specified value and utilization of dither to decrease high friction of harmonic drive gears. Moreover, we propose to use a mechanical shock absorber that consists of a contact part with small mass and a proper stiffness and damping property for when the impact force cannot be neglected in the contact. The experiments based on the proposed method show that the magnitude of the contact force can be limited to around 10 [N] while the performance of the position control can be also adequate.
In this paper, we describe development of a small six-component force/torque sensor based on the double-cross structure. First, we propose the double-cross structure that consists of serially connected parallel plate structures. Second, we explain how to use the parallel plate structure as a two-component force/torque sensor element that detects a force and a moment. Then the proposed force/torque sensor is shown to be designed as a six-component force/torque sensor with the ideal force sensitivity. A design detail of this sensor is explained and its prototype is shown. Finally, experimental results illustrate the performance of the prototype and the effectiveness of the proposed double-cross structure.
In this paper, we deal with an optimization of grasping when we not only resist an external force applied to a grasped object but also generate a desirable acceleration of the object. Based on the concept of required external force set, we define required acceleration and equilibrium-force sets. By using the sets, we define an optimization problem from the viewpoint of decreasing the magnitudes of the joint torques required to generate the acceleration and the equilibrium-force, and show that we can solve the problem by using a branch-and-bound method. We also present numerical examples in order to show the validity of our approach.
Presentation of haptic sense or sense of force makes the feeling of being at the scenes enhancing in a virtual reality world. Most of the force display systems being into practical use are, however, small size and do not have enough large generative force and working area for many applications such as rehabilitation therapy, skill training, designing, game or so. On the other hand, a force display system is a kind of robot, which operates in constant contact with human beings, so it must be safe in all operational conditions. In this paper, we discuss the safety properties of a middle- or large-sized force display system using clutches whose input shafts are driven at low speed, and then consider their force display characteristics.
This paper describes an omni-directional power assisted cart with a collision avoidance function. At each four corner of the vehicle, we installed two ultrasonic sensors, one directed to the side of the vehicle and on along its diagonal. There are three features. (1) The distance sensed by the front-side ultrasonic sensors is memorized. As the relative displacement of the cart can be calculated, it is possible to incrementally built-up the topology of the surrounding environment of the cart and thus follow the position of obstacles even in the areas where no ultrasonic sensors have been installed. (2) It is possible to set up the position of the rotation center of the vehicle by adapting its control laws. Collision with obstacles in the in-course and out-course of the vehicle can be avoided by shifting the rotation center. (3) The collision can be avoided in the forward, backward and side directions by, first, slowing down the cart and then eventually stopping it. These three capabilities enabled us to reduce the number of ultrasonic sensors to the minimum necessary (8 pieces) ; and also to ensure comfortable conditions of use to the user even if one of the sensors performs a wrong measurement.
This paper discusses the pushing manipulation of an object by a humanoid robot. We generate the walking pattern for the pushing task by modifying the desired ZMP trajectory of the walking pattern for a humanoid robot whose hands do not contact with an object. For the modified walking pattern, we analyze whether or not the robot can keep the dynamical balance by considering both the ZMP and the walking velocity. The validity of the proposed method is confirmed by some simulation results. We further consider extending the proposed algorithm by separating the walking phase and the pushing phase.
A leg-wheel robot has mechanically separated legs and wheels, and it performs high mobility and stability on rough terrains. In the first part of this paper, we propose a gait algorithm, which allows continuous locomotion of the robot under the random velocity command by a human operator. The gait algorithm, which uses predictive control, determines the timing of the legs' lifting to avoid the legs reaching the border of the work space. The predictive control is based on the comparison between the time necessary to return the support legs and the time left before they reach the border of the work space. In the second part, a velocity limitation method for the gait algorithm is discussed. This method limits the velocity command when it exceeds the mechanical specifications of the robot. Combined the velocity limitation method with the gait algorithm, it ensures the continuity of locomotion, and gives the efficient gait pattern with a long step length and low frequency of leg phase change. The proposed algorithms are evaluated by simulation and experiments on rough terrain.
The authors developed a prototype hexapod with integrated limb mechanism, named“MELMANTIS-1.” MELMANTIS-1 uses a six-bar linkage with four degrees of freedom as an example of integrated limb mechanism with features fittable to both of leg and arm. This paper describes the multimodal control of MELMANTIS-1 for legged mobile manipultion. Seven control modes are developed for MELMANTIS-1; hexapod omni-directional walk on flat floor, hexapod omni-directional walk on rugged terrain, transformation between leg posture and lift-upped arm posture, transformation of configuration with the body and legs between radially symmetric hexapod and side-by-side symmetric quadruped with dual arm, object manipulation by dual arm, quadruped walk with handling an object, and individual joint control of single leg. All control modes are commanded by a human operator's numerical keypad input. Human operator can select the control modes depending on the situation of robot and can achieve various tasks. Especially, the configuration change of the body and legs, between radially symmetric hexapod and side-byside symmetric quadruped with dual arm, is strikingly effective to enhance the working performance of multi-legged robot including both of mobility and manupulability.