This technology converts human facial images into grayscale vectors, reduces dimensions via PCA, and numerically calculates the arousal and valence values of the A-V emotion model using linear regression analysis.
Existing emotion classification methods only recognize discrete, categorized basic emotions, which limits continuous human-robot interaction due to facial tracking failures or recognition errors.
By combining PCA with a linear regression model, this technology numerically estimates coordinate values in the A-V emotion space from continuous video frames in real time.
This technology is an automated inspection mechanism that utilizes a body traveling between rails equipped with multiple non-contact optical sensors to measure rail gauge (width), surface condition, notches, edge damage, and cracks at connection points, with a control unit that analyzes this data to provide maintenance insights.
Conventional rail inspection is performed manually by workers, which is time-consuming and labor-intensive, while also being prone to subjective judgment, missed inspections, and safety risks in the work environment.
This technology implements an automated system that precisely measures the condition of each rail section using first through fifth sensors positioned on both sides and the top guide of the body. It utilizes cameras for autonomous navigation and a control unit for data analysis to identify and report damage locations to operators. Applicable to logistics transport, service robots, and autonomous platforms, it enhances the efficiency and accuracy of rail inspections, reduces the risk of accidents and human error, and minimizes the need for manual labor and associated costs.
This invention was developed with support from the Ministry of Education, Science and Technology for the development of convergence technology for new and renewable energy intelligent robots.
This technology combines direct operator teaching with automated robot control assistance for peg-in-hole assembly processes involving multi-peg components. It optimizes robot playback performance by automatically or selectively removing teaching data from unnecessary segments where no robot movement occurs, using linear and angular velocity analysis.
Conventional position-based direct teaching is difficult to implement for complex assembly tasks involving contact, such as inserting multiple pegs. Furthermore, inefficient stationary data generated during manual teaching by operators often leads to unnecessary delays in robot playback time.
This technology establishes a teaching procedure for inserting multiple pegs of varying lengths in stages (primary and secondary). It applies a data editing algorithm that analyzes the linear and angular velocity components based on the robot's tool coordinate system to identify stationary segments, then automatically or selectively removes that data via a user interface. Applicable to robot gripping, precision measurement, and automated equipment, it improves overall assembly efficiency and reduces robot playback time.
This technology calculates the position and orientation of a moving object by applying just two feature points extracted from a 2D camera image and their corresponding 3D information into a perspective projection matrix equation. It uses trigonometric synthesis to compute rotation (sinθ, cosθ) and translation (tx, ty) data at high speeds.
Conventional visual odometry technologies require a large number of image points to ensure accuracy, leading to high computational loads and limitations in real-time processing due to iterative optimization. They also rely heavily on expensive IMU or GPS/INS sensors, resulting in low cost-efficiency.
This technology defines a perspective projection matrix equation using only two image points and derives rotation and translation data using internal parameters (focal length, principal point). In particular, it significantly reduces computational complexity by synthesizing expressions containing sinθ and cosθ into a single trigonometric function.
This technology performs image acquisition based on stereo vision and motion detection using a block-matching algorithm. It improves the precision of object motion estimation by scaling motion vectors in consideration of the PTZ camera's zoom magnification, and controls the robot's response behavior based on the detected event situation.
Conventional single-camera or pan-tilt camera systems suffer from blind spots in complex environments and are unable to perceive 3D hazards due to their reliance on 2D imagery. Existing motion detection based on frame differencing is prone to noise and false positives, while optical flow methods are often unsuitable for real-time processing due to high computational requirements.
This technology acquires 3D distance information through stereo vision using multiple cameras and tracks motion by implementing a block-matching method in the event detection unit. In particular, it includes a correction logic that scales motion vectors according to changes in camera zoom magnification, ensuring accurate object motion detection even in variable shooting environments.
This technology consists of a multi-jointed robotic arm located at the center of the upper body, along with peripheral tableware and tilting components. It is a mechanism designed to assist patients with severe disabilities in eating through the robotic arm's scooping motion and a rack-and-pinion-based container tilting control.
Patients with severe disabilities face physical limitations in eating independently without the help of a caregiver, which leads to psychological issues such as a decline in self-esteem.
This technology features a servo-motor-based multi-jointed robotic arm that scoops and serves food, while a tilting adjustment component equipped with a rack-and-pinion mechanism at the base adjusts the angle of the food container based on the remaining amount, ensuring the user can consume every last bit.
This technology is a system and control algorithm for remotely operating a toy-style autonomous robot via a short-range wireless communication (Bluetooth) software application on a user device. It includes a structure that allows discarded smartphones to be integrated with the robot body, repurposing them as a camera and control interface.
Existing toy robots are primarily focused on simple movement or cleaning functions, and lack effective remote control interfaces using user devices or integrated service platform environments.
This technology establishes a communication channel between a server, a user device, and a toy-style robot. It provides a remote control method that enables robot movement, camera control, and resolution settings through a process of downloading programs from the server and authorizing Bluetooth permissions.
This technology features a multi-jointed mobile robot that connects its body and arms using multiple link members, with curved sections and elastic damping structures formed on the underside of each, allowing it to adaptively navigate and traverse stair steps.
Conventional wheeled robots are efficient for travel on flat surfaces with uniform height, but they face structural limitations when effectively navigating uneven terrain such as stairs.
This technology incorporates repeating curved sections along the bottom of the housing and arms, utilizes elastic damping members, and controls relative positioning via rod-shaped connecting members to vary the contact surface. It can be applied to indoor delivery robots and building inspection robots, enabling smooth and stable movement even in environments with stairs and thresholds.
This technology features a vibratory robot that forms a tubular structure by connecting multiple vibration modules in a ring shape. Each module contains an electromagnetic vibration unit composed of a coil, a magnet, and an elastic member, allowing for the control of vibration amplitude, frequency, and phase to achieve both flight and driving locomotion.
Existing flying robots are significantly affected by air currents and have limited low-altitude flight capabilities, while ground-based robots face movement constraints due to the physical limitations of wheels or tracks, which restrict their range depending on terrain conditions.
This technology changes its structure between flight and driving modes by varying the angles of the top and bottom of the modules. By applying individual AC power to each module, it controls thrust and direction through asymmetric amplitude and phase modulation. Applicable to indoor exploration and disaster site reconnaissance, it introduces a new mode of mobility that allows a single robot to transition between flying and driving.
This technology calculates the distance between a mobile robot and a smartphone by measuring the RSSI (Received Signal Strength Indicator)-based path loss between Wi-Fi transmitters attached to three or more robot arms and a smartphone receiver, and estimates the robot's self-position through triangulation and geometric calculations.
Existing indoor positioning technologies are inefficient in terms of resources and time, as they require the construction of expensive, dedicated embedded platforms and prior knowledge of node locations.
This technology rotates the mobile robot to align the relative angles between the robot arms and the smartphone, then applies the Wi-Fi RSSI-based Friis transmission equation and triangulation to calculate the position in real-time on the smartphone platform. It can be applied to logistics transport, service robots, and autonomous driving platforms, thereby improving the efficiency of robot localization without additional infrastructure and reducing resource waste by utilizing the smartphone platform.
This invention was developed with support from the Ministry of Education, Science and Technology for the development of convergence technology for new and renewable energy intelligent robots.
This technology is a proximity control mechanism that determines driving direction and guides docking positions by detecting the RGB-LED brightness of a target robot to ensure precise coupling between modular robots. By arranging three-color RGB-LEDs and sensors radially, it provides location information and guides the docking range of the target robot.
Conventional ultrasonic sensors are limited to obstacle avoidance and lack the precision required for accurate positioning, while RF signal strength (RSSI) methods suffer from low recognition accuracy and errors when docking moving objects.
This technology features a sensor module (comprising three RGB-LEDs and one detection sensor) arranged radially around the robot's body. It converts the RGB-LED brightness values of the target robot into frequencies, generates a driving path toward the direction of maximum frequency intensity, and implements a control algorithm to stop at the target docking point. Applicable to logistics, service robots, and autonomous platforms, it enhances the accuracy and efficiency of modular robot bonding across various applications.
This invention was developed with support from the Ministry of Education, Science and Technology's New and Renewable Energy Intelligent Robot Convergence Technology Development program.
This technology is a finger rehabilitation device and system that secures the user's palm to a support plate and inserts the fingertips into individual wearing units, converting motor rotation into finger movement through a hinge link structure and gear coupling.
Conventional Velcro-based glove structures are difficult for patients to put on independently, and wire-driven systems suffer from control precision issues due to material deformation (stretching) and interference with thumb movement.
This technology features a gear-driven unit where sector gears and rotating gears mesh for the four fingers (excluding the thumb), and a secondary drive unit that enables 2-axis (vertical/horizontal) movement for the thumb, allowing for precise range-of-motion and rotation control for each finger.
This technology is a mobile robot system that controls driving modes (following/leading) by combining LF signal-based triangulation with gait status measurement from a wearable device. It also performs infrared-based alignment between multiple robots and gesture control via smartwatch.
Existing image processing methods require high-performance processors and consume significant battery power, while sensor network methods are limited to specific spaces and struggle with precise positioning and navigation in complex, obstacle-filled environments.
This technology utilizes multiple LF transmitters on the mobile robot to perform triangulation based on signal strength received by the user's wearable device, enabling precise positioning. It automatically switches driving modes (Mode 1: Following, Mode 2: Leading) based on the user's gait and rotation, and includes alarm and gesture control via smartwatch, as well as formation driving control through infrared communication between multiple robots.
This technology is a tilt-rotor multicopter mechanism that connects some of the propellers arranged radially on the airframe to a robotic arm-style tilting motor unit, allowing the rotation angle of the propellers to be variably controlled between vertical (for takeoff and landing) and horizontal (for high-speed flight) depending on the flight state.
While conventional rotary-wing drones are capable of vertical takeoff and landing, they must tilt significantly during high-speed flight, creating a trade-off between maintaining stability and achieving high speeds when transporting cargo.
By equipping some of the six propellers with Dynamixel (robotic joint motors) and tilting arms to control the angle of the fixed motor units, this technology maintains vertical takeoff and landing capabilities while maximizing horizontal thrust to achieve speeds comparable to fixed-wing aircraft.
This technology models an object's motion data using NURBS (Non-Uniform Rational B-Splines) equations to generate multiple curves that share a time axis, and controls the motion of the object by adjusting control points and weights to regenerate and synchronize the motion.
Conventional spline-based motion modeling is vulnerable to measurement noise and makes it difficult to modify or precisely control modeled curves for specific purposes.
This technology builds a NURBS-based model from motion measurements and applies parameter increment calculations and interpolation to adjust control points and weights in real-time, enabling precise control of the object's position, velocity, and torque while maintaining a shared time axis.
