A fundamental idea for constructing a conscious robot is presented. First, hypotheses of the human mind are presented. The following ideas are shown: (1) The unconscious system is a recurrent network system made of various distributed subsystems. (2) Information of the mind such as intellect, feelings and willpower, is presumably processed in the unconscious system instead of the conscious system. (3) The conscious system just monitors, experiences afterward, models and memorizes the results of the unconscious system. (4) Realistic experiences of quality that the conscious system feels by itself are just illusions that are defined in the brain. Then an algorithm of a robot mind is constructed based on the hypotheses mentioned above. It is shown that a conscious mind of robots can be made using the proposed algorithm. Finally, purposes and issues of the robots with a mind are also discussed.
This study has been conducted by participating in the Humanoid Robotics Project (HRP) of the METI, which aims at developing applications of humanoid robots. We consider the humanoid robot is effective in applications that covers the services to human beings such as elderly care, patient care, since it presents affinity to people owing to its human-like figure and motions. In this study, our focus was a situation in which the robot is expected to support activities of people in a hospital with three kinds of possible system users who are a nurse, a patient and a family member in the remote site. Taking the capabilities of current humanoid robots into account, we designed the user interface of the system which satisfies the requirements of the each three users to operate the robot in the assumed situation. Furthermore, we confirmed the feasibility of the designed system through experimental examples by integrating the developed user interface with a humanoid robot as a total system.
This paper describes a view-based outdoor navigation method. In the method, a user first guides a robot along a route. During this guided movement, the robot learns a sequence of images and a rough geometry of the route. The robot then moves autonomously along the route with localizing itself based on the comparison between the learned images and input images. Since appearances of objects in images may vary much according to changes of seasons and weather in outdoor scenes, a simple image comparison does not work. We, therefore, propose a comparison method in which the robot first recognizes objects in images using object models which allow for such appearance variations, and then compares recognition results of learned and input images. A method is also developed which automatically selects key images used for the comparison from an image sequence. Experimental results of about 350 [m] autonomous navigation in our campus under various conditions show the feasibility of the method.
This paper proposes a method of pose (position and orientation) estimation of objects using multiple ID devices attached to each object for autonomous robots. The object pose is estimated from geometrical relation between poses of the devices attached to the object and the reading poses of the reader. It is required the poses of at least two ID devices in different orientations. We propose a method of precise localization of the ID device even when the ID reader has the volume of communication area. We show the feasibility of our proposed method through the pose estimation experiment.
A newly designed step climbing wheeled robot is discussed with detailed analysis. The proposed mechanism has a compliant element between the front and rear wheels to help the wheels climbing the step which is higher than the wheel radius. The analysis gives the optimal compliance of the compliant element to keep the contact between the wheel and the step wall as long as possible. Some simulation and experimental results are also given to show the effectiveness of the proposed mechanism.
For real-time decision making of a robot, there is an approach that utilizes the look-up table of the pre-computed result of dynamic programming. The look-up table records appropriate behavior for every situation of the robot and its surroundings. A robot that is installed the look-up table can decide its behavior only with a reference of the table. However, a table is usually too large to be loaded on the memory of usual robots. For the solution of this problem, we have proposed to use vector quantization for compressing the table. In this paper, we evaluate this method quantitatively. Then, we newly introduce an information entropy function that searches an appropriate way of blocking. For simulations and experiments, a look-up table for soccer behavior was created and compressed. As the results, the entropy function could find an appropriate way of blocking and the compression method with the blocking way enable the table to be compressed to 1.5% size.
This paper describes the development of a high speed dynamic simulator for humanoid robots. In the simulator, an order n formulation is used to solve the inverse dynamics and forward dynamics of a multi-body system. The formulation can deal with a tree structure and multiple contacts with the environment. In order to simulate a collision with friction between the bodies and environments, a virtual spring-damper contact model is proposed. This model enables an accurate computation of the reaction forces and slips. A simulation of ascending steep stairs is carried out in order to demonstrate the validity of the simulation. The results of the simulation are presented and discussed.
In this paper, we propose a new “slip-adaptive” strategy which enables a robot to walk on slippery surface. In this strategy, the gradient of the surface is estimated based on the acceleration signals measured through the accelerometer attached to each leg. A short-time force control considering the friction cone is applied to the slipping leg for slip compensation. This input consists of the position/force hybrid control of the supporting legs (i.e., a position control of kicking motion and a force control of slipping compensation) . The validity of the proposed method is confirmed through the simulation and the experiment.
This paper studies an influence of the foot shape on energy-efficiency of passive dynamic walking of a bipedal robot. Two types of bipedal robots are compared in terms of variation of kinetic energy under locomotion. One has point-feet, and another has linear feet. Under the simplification of the robot models, it is proven that the robot with linear feet presents larger walking speed in steady walking compared to the point-footed robot. Is is also verified by numerical simulations that the above claim holds for the case of the original (unsimplified) model. A design method of foot shape is proposed.
In this paper, we propose a decentralized control algorithm of multiple mobile manipulators handling a single object in coordination. In this algorithm, the grasping point of each robot is controlled as if it has a virtual 3-D caster dynamics and the handling of a single object by multiple mobile manipulators is realized based on a leader follower type control algorithm without using the geometric relations among robots. The proposed control algorithm is experimentally applied to multiple mobile manipulators, and the validity of the proposed control algorithm is illustrated by the experimental results.