Abstract: A walking robot and a control method thereof. The robot includes at least one joint unit on each leg, a sensing unit to sense angle and angular velocity of the at least one joint unit, a memory unit to store data of the angle and angular velocity during stable walking, a target trajectory generation unit to generate a target trajectory, a control torque calculation unit to check stability of the at least one joint unit by comparing the sensed angle and angular velocity with the target trajectory, and, if an unstable joint unit is present, to calculate a control torque of the unstable joint unit to trace the target trajectory, and a servo control unit to transmit the calculated control torque to the unstable joint unit, thereby controlling torques of joint units using an FSM without solving the complicated dynamic equation, thus achieving stable walking.

Abstract: An object recognition system including an image data storage unit to store a captured image, a feature extraction unit to extract an image having a predetermined rotational component among rotational components of the image stored in the image data storage unit and to extract feature vectors based on the extracted image, a database unit to store object information, a recognition unit to determine whether an object corresponding to the captured image is present in the database unit through comparison between the feature vectors extracted by the feature extraction unit and the object information stored in the database unit and to recognize information on the object stored in the database unit based on determination as to whether the object corresponding to the captured image is present in the database unit, and a system administration unit to receive the information on the object recognized by the recognition unit.

Abstract: A torque-based walking robot and a control method thereof which stably controls walking of the robot. In the control method, in which high rigidity, equal to that achieved through a position-based control method, is achieved using a torque-based control method without switching between the position-based control method and the torque-based control method while the robot is in motion, a difference between a target torque and a measured torque is forcibly generated by limiting a torque range measurable by each torque sensor, thereby increasing voltage applied to each actuator, and thus achieving high rigidity, equal to that achieved through the position-based control method, using the torque-based control method without switching between the position-based control method and the torque-based control method.

Abstract: A moving robot apparatus and a control system and method thereof. A current state variable (e.g., the current direction of the main body of the robot apparatus) may be estimated using current data output from a sensor part of the robot apparatus. Initially, it is determined whether the current data output from the sensor part satisfy a necessary condition. The necessary condition may be considered to be satisfied when the current data are unaffected by a disturbance or external interference (e.g., the earth's magnetic field.). Thereafter, a current sample value may be extracted from a previous sample value obtained from the sensor part. The previous sample value may relate to a previous state variable (e.g., a previous direction of the robot main body). A weight of the current data may be calculated based on the current sample value. A value of the weight may depend on whether the necessary condition is satisfied by the current data.

Abstract: A compliant joint capable of achieving passive compliance for a robot in order to prevent an injury to a human by collision with the robot. The compliant joint includes a housing, a cam member rotatably mounted in the housing, a roller spring device mounted to the cam member to be compressed and extended, a guiding member formed in the housing to guide compression and extension of the roller spring device in accordance with rotation of the cam member, and a receiving recess formed at the guiding member to engage the housing and the cam member with each other by receiving the roller spring device and release the engagement by separating from the roller spring device. Accordingly, robustness of the robot can be maintained when an impact less than a predetermined magnitude is applied, while being suddenly decreased when an impact greater than the magnitude is applied.

Abstract: A method of extracting a three-dimensional (3D) edge is based on a two-dimensional (2D) intensity image and a depth image acquired using a time of flight (TOF) camera. The 3D edge extraction method includes acquiring a 2D intensity image and a depth image using a TOF camera, acquiring a 2D edge image from the 2D intensity image, and extracting a 3D edge using a matched image obtained by matching the 2D intensity image and the depth image.

Abstract: A walking control apparatus of a robot includes a joint portion provided in each of a plurality of legs of the robot, a pose sensing unit to sense the pose of the robot, a walking state determination unit to determine a walking state from the pose of the robot, a knot point compensation value calculator to determine a Center Of Mass (COM) of the robot from the pose of the robot and to calculate a knot point compensation value, a desired angle trajectory generator to generate a reference knot point of the joint portion corresponding to the walking state, to compensate for the reference knot point using the knot point compensation value so as to generate a desired knot point, and to generate a desired angle trajectory of the joint portion using the desired knot point.

Abstract: A robot and a control method thereof may adjust a yaw moment generated from a foot contacting a ground to achieve stable walking of the robot. The robot, which may have an upper body and a lower body, may include a main controller starting walking of the robot through only motions of joints of the lower body and adjusting a motion of the upper body such that a yaw moment generated from a foot the lower body during walking of the robot is less than the maximum static frictional force of a ground to perform stable walking of the robot, and sub controllers driving actuators of the joints according to a control signal of the main controller.

Abstract: A manipulator, and a control method thereof, calculates a degree of freedom (DOF) used to prevent collision with an obstacle in consideration of an order of priority of DOFs for an operation and collision avoidance contribution. The manipulator judges whether there is danger of collision of the manipulator with an obstacle, judges whether or not there is at least one operating DOF capable of avoiding collision with the obstacle among the plurality of operating DOFs, upon judging that there is danger of collision with the obstacle, and avoids collision with the obstacle using at least one operating DOF having the lowest priority while performing the operation among the at least one operating DOF capable of avoiding collision with the obstacle, upon judging that there is the at least one operating DOF capable of avoiding collision with the obstacle.

Abstract: A walking control apparatus of a robot includes joint portions provided in each of a plurality of legs of the robot, a state database to store state data of each of the legs and state data of the joint portions corresponding to the state of each of the legs, when the robot walks, a position instruction unit to store desired positions corresponding to the state data of the joint portions, an inclination sensing unit to sense an inclination of an upper body of the robot, a torque calculator to calculate torques using the inclination of the upper body and the desired positions, and a servo controller to output the torques to the joint portions to control the walking of the robot. Since the robot walks by Finite State Machine (FSM) control and torque servo control, the rotation angles of the joint portions do not need to be accurately controlled. Thus, the robot walks with low servo gain and energy consumption is decreased.

Abstract: Disclosed herein are an apparatus and method for controlling stable walking of a robot based on torque. In a method of enabling stable walking by controlling torque of a hip joint portion using a Finite State Machine (FSM) without solving a complicated dynamic equation, torque of a stance leg is finally calculated using pose control torque of an upper body, pose control torque of a swing leg, and initial pose control torque of a stance leg supporting the upper body. Accordingly, the robot may stably walk with torque balance. Since gravity compensation torque is applied, a torso of the robot is not inclined and the pose of the robot is stably maintained.

Abstract: An image-based localization feature point registration apparatus includes a camera to capture an image, a feature point extractor to extract a feature point from the captured image, a calculator to calculate depth information about the feature point according to whether the feature point is one of two-dimensional (2D) and a three-dimensional (3D) corner, and a feature point register to register 3D coordinates of the feature point based on the depth information about the feature point and image coordinates of the feature point.

Abstract: An augmented reality (AR) service apparatus includes a camera to capture an actual image, a controller to receive feature point information about the captured image from at least one of a plurality of base stations (BSs), detect a location of the camera by matching data of feature points with data of the image, and provide location-based information in a same direction as the captured image according to the location of the camera, and a display to realize an AR service by combining the captured image with the location-based information under control of the controller.

Abstract: Two-dimensional image information and three-dimensional image information of a subject are acquired, facial recognition is performed using the two-dimensional image information to determine whether a recognized face is a registered user's face, an elliptical model of the user is matched to the three-dimensional image information to calculate an error if it is determined that the recognized face is the user's face, and it is determined whether the user's face is improperly used based on the error. The subject's face is determined using the two-dimensional image information and the three-dimensional image information of the subject and it is determined whether the recognized face is improperly used, thereby improving facial recognition reliability. Thus, information security is improved.

Abstract: A robot one of transfers an article to an external subject and receives the article from the human subject and flexibly copes with changes in human intention. If a pushing force applied to robot hands is sensed, the robot hands grip and pull the article to inform the external subject that the robot hands are prepared to receive the article from the external subject, and transfers the article to the external subject or takes the article from the external subject according to whether the pushing force or a pulling force applied to the robot hands is sensed. If the pulling force applied to a robot hands is sensed, the robot pushes the article to inform the external subject that the robot hands are prepared to transfer the article to the external subject, and transfers the article to the external subject or takes the article from the external subject according to whether the pushing force or the pulling force applied to the robot hands is sensed.

Abstract: An apparatus, method and computer-readable medium planning a path of a robot by planning an optimal path while satisfying a dynamic constraint. In a process of searching for a motion path from a start point to a goal point while extending a tree from a start point of a configuration space to generate a path, along which a manipulator of the robot is moved in order to perform a task, an optimal path is generated responsive to the dynamic constraint of the manipulator of the robot to generate stable motion satisfying momentum and Zero Moment Position (ZMP) constraint. Accordingly, path planning performance is improved and a path satisfying a kinematic constraint and a dynamic constraint is rapidly obtained.

Abstract: A path planning apparatus of a robot smoothes a motion path while satisfying a constraint. In a configuration space where a manipulator of a robot performs a task, a Rapidly-exploring Random Tree (RRT) path which extends from a start point and reaches a goal point may be smoothed while satisfying a constraint to generate a stable motion path. Accordingly, path planning performance is improved and an optimal path satisfying a kinematic constraint may be rapidly obtained.

Abstract: A robot and a method of controlling the same are disclosed. The robot derives a maximum dynamic performance capability using a specification of an actuator of the robot. The control method includes forming a first bell-shaped velocity profile in response to a start time and an end time of a motion of the robot, calculating a value of an objective function having a limited condition according to the bell-shaped velocity profile, and driving a joint in response to a second bell-shaped velocity profile that minimizes the objective function having the limited condition.

Abstract: Disclosed herein are a humanoid robot and a control method thereof. The humanoid robot controls robot parts performing main motions having high relevance with respect to a commanded action to the humanoid robot such that these robot parts move along optimized motion trajectories generated through motion optimization in consideration of robot dynamics, and controls robot parts performing remaining motions having low relevance with respect to the commanded action such that these robot parts move along predetermined motion trajectories corresponding to the commanded action, thereby simplifying optimization of whole body motions of the humanoid robot while performing the commanded action maximally similarly to a real human action.

Abstract: Disclosed herein are a markerless augmented reality system and method for extracting feature points within an image and providing augmented reality using a projective invariant of the feature points. The feature points are tracked in two images photographed while varying the position of an image acquisition unit, a set of feature points satisfying a plane projective invariant is obtained from the feature points, and augmented reality is provided based on the set of feature points. Accordingly, since the set of feature points satisfies the plane projective invariant even when the image acquisition unit is moved and functions as a marker, a separate marker is unnecessary. In addition, since augmented reality is provided based on the set of feature points, a total computation amount is decreased and augmented reality is more efficiently provided.