A legged robot “JROB-1” is developed as a platform for the research on perception-action coupling in intelligent behaviour of robots. The JROB-1 has the features as follows: 1) robot base is TITAN-VIII, 2) PC/AT compatible on-board computer which is controlled by RT-linux, 3) self-contained and connected to a network via radio ethernet, 4) many commercial boards (including color tracking vision board) are available, 5) extensibility with respect to hardware and software, and 6) every parts is commercially available. JROB-1 is one of the accomplishment of Grant-in-Aid for Scientific Research on Priority Areas by Ministry of Education, and is expected to be a common test-bed in experiment and discussion for various aspects of intelligent robotics.
The purpose of this research is to make a machine or a robot jump highly using small output actuators. It is sometimes necessary for a robot to jump over a big obstacle. This usually requires large output actuators. But it is difficult to find light actuator with large output. A jumping machine with a self-energizing spring system is introduced, which is designed to jump using small actuators and springs. Power for the self energizing system is generated using small actuators. And it can jump highly. Computer simulations and experiments are performed to prove the validity of the jumping machine and the self energizing system.
We have developed a task planning system for an autonomous robot to maintain equipment in hazardous environments such as nuclear power plants. Generally, maintenance tasks are semi-structured tasks, and a model-based robot system is used for their execution. However, it is difficult to make exact environment models constantly at a work site. Therefore, we have developed a practical task planning system which can perform desired tasks smoothly, even though environment models include some errors. In this system, state value named ‘sureness’ is introduced to indicate the amount of error in the environment models. By using this value, a better procedure can be selected from the task knowledge database for performing a task. In this paper, we describe a task planning method with sureness value for the autonomous maintenance robot system, propose a procedure to define sureness value using fuzzy mem-bership functions, and present a developed task planner for assembling a nut and a bolt to a flange based on above task planning framework. We verified that suitable robot control programs were generated from the operator's task command, and assembling task of a nut and a bolt to flange were performed autonomously by using the developed task planner.
We have proposed an omnidirectional image sensor called HyperOmni Vision. This sensor can acquire omnidirectional and perspective images in real time. In this paper, we propose a new Hough transform method for an omnidirectional image by using a cubic Hough space. Furthremore, This report presents a method for reconstructing 3D line segments from an omnidirectional input image sequence obtained by HyperOmni Vision with known motion.
A spatial path generation algorithm based on visual sensory data and its application to the seam tracking robotic system which can operate even under poor sensing conditions are presented. At first, the concept of the robotic system with a wrist-mounted high performance vision sensor is presented, then an accurate spatial path generation algorithm using sensory data with a reliability coefficient is described. In the algorithm, piecewise polynomial functions are generated from sensory data that is acquired ahead of an end-effector, and the spatial path is recovered by connecting the piecewise polynomial functions successively under the minimum error condition as well as the boundary conditions. Computer simulations and experiments on a spatial path on a curved surface show that the proposed system operates effectively even under disturbances such as sensing noise, irregular tack-welding-beads and a burst of lack of sensory data, often appearing in practical applications.
This paper proposes an approach to analyzing and designing an intelligent vehicle controller for partially supporting the driver's operation of a vehicle. In the process of driving a vehicle, the driver has certain expectations of vehicle behavior. The driver operates the steering wheel and accelerator and brake pedals in an effort to achieve those expectations as the vehicle moves along in the driving environment. Moreover, the driver perceives the vehicle's spatial movement and determines the next driving operation based on the relationship between vehicle behavior and the expectations of that behavior. Thus, driving can be thought of as a system formed by the interaction between the driving environment, vehicle behavior and the driver's expectations of vehicle behavior. This paper proposes a model of driving environment, driver and vehicle behavior interaction as a tool for analyzing and designing a vehicle controller and vehicle control characteristics adapted to the driving environment and the driver's intentions. This interaction model incorporates transitions representing knowledge of the driver's particular cognitive characteristics. These transitions are expressed using an extended Petri net description method and are adopted among the model rules describing driver behavior. The Hierarchical Fuzzy Integral (HFI) is used as a multipurpose decisionmaking technique that allows explicit treatment of the driving environment, vehicle behavior and the driver's intentions. The characteristic of the driver's cognition of the driving environment is treated as affordance. Based on the affordance perception between the driver and the driving environment, the differences in vehicle behavior demanded by individual drivers have been expressed in engineering terms by varying the fuzzy rules of HFI. As an example, a control procedure designed with the proposed model is applied to automatic engine braking control during downhill coasting. The simulation results show good agreement with driving test results.
This paper describes a learning system that has the ability to estimate the flight trajectory of a spinning ball. This system plays an important role in realizing a robot that plays ping-pong against a human. LWR (Locally Weighted Regression) is employed to learn a forward map from stereo image inputs of an incoming ball to the state outputs of the ball just before hitting. The experimental results show that the ball trajetory can be estimated accurately irrespective of how the ball is spinning.
We have designed several omnidirectional image sensors such as COPIS (COnic Projection Image Sensor) with a conical mirror and HyperOmni Vision with a hyperboloidal mirror, for navigating the robot in an environment. Described here is a method for generating an environmental map by using the Omnidirectional Image Sensor. Based on the assumption of translational motions of the robot, the proposed method can generate an environmental map and estimate egomotion of the robot. In particular, the method is suitable for map generation during long robot movement because deadlocking error of the robot does not accumulate. The method has been evaluated on a prototype of HyperOmni Vision in actual environment.
This paper proposes a Contact Interaction Robot (CIR) and reports CIR's psychological effect to the human. The CIR utilizes contact behavior as the interaction means between a human and the robot. We develop the CIR equipped with pressure sensors on both sides of its neck and six servo motors in its neck, two arms and two legs taking account of the application to the dentist as a typical example. The psychological experiments are performed by utilizing the CIR. The experimental results reveal that the CIR is able to moderate the painfulness perceived by the human as well as to bring a sense of relief.
This paper proposes a monitoring system for a human respiration and posture in sleep using pressure sensor array. The proposed system consists of 221 pressure sensors (Force Sensing Resistors: FSRs) attached to the surface of the bed. Each sensor of the pressure sensor array is set 5 [cm] apart. The novel features of the proposed system lie in non-invasive and unrestrained monitoring of the human respiration and posture. Non-invasive monitoring eliminates the need for monitoring needles or catheters to invade the human body, sensors thus do not impose a physiological burden such as pain on him or her. In unrestrained monitoring, sensors and their electrical cords do not limit degrees of freedom of his or her movement. Unrestrained sensing therefore does not impose a psychological burden caused by the limitations on him or her. Experimental results demonstrate that the proposed system is feasible for monitoring a human respiration and posture for over 6 hours.
While there have been a number of works on enveloping grasp, most of them have discussed the robustness of grasp, analysis of contact forces, and localization of contact points, under the assumption that robot fingers already grasps the object. This paper relaxes the assumption and treats an issue on planning torque commands for lifting up the object on the table till it makes contact with the palm. We first define a concept of transition stability during lifting motion and introduce the force-flow-diagram which is a convenient tool for evaluating the global moving direction of object. We show a couple of simulation results for both 2D and 3D objects.
In this paper one deals with the inverse kinematics of serial-link manipulators with redundant DOFs. At first, one proposes a new algorithm to solve the joint rate variables of redundant manipulators in the framework of the weighted generalized inverse of Jacobian. In the proposed algorithm the weights are determined so as to avoid the computational illconditioning during the recursive procedures of computation, which means no off-line task for determining weights is necessary. Some ways of computing the weights are considered and evaluated with respect to condition number in the simulation. The proposed algorithm takes the least additional load to compute the projection operator which projects arbitrary joint rates into the null-space of the end-effector's Cartesian coordinates. It is due to the algorithmic structure of the proposed algorithm in which some variables produced in the process of the generalized inverse of the Jacobian are made use of computing the projection operator. The anthropomorphic type seven DOF manipulator was taken as an example to evaluate the proposed algorithm. Some ways to determine the arbitrary joint rate vectors were considered so as to make the manipulator behave like a human. Simulations also reveal the stable behavior of the manipulator due to the proposed algorithm when the manipulator is indicated to take singular postures.
Positioning system of vehicle on undulating area is developed. The system uses a fiber optic gyro, roll pitch sensor and two wheel encoders. The system realizes high precision positioning by compensating the effect of undulation using roll pitch sensor data. A prototype of autonomous mower equipped with the positioning system was produced and its performance was evaluated. The results showed that this positioning system compensated the effect of undulation and its positioning performance was equal to the performance on flat area.