Byung-jin Jung

Collision detection using band designed Disturbance Observer

In this work, we design frequency band filters in Disturbance Observer (DOB) for collision detection between the industrial robot manipulator and human. We realize that the frequency characteristics of the inverse dynamics of the manipulator is similar to that of high pass filter and the sensor noise filter is in the form of low pass filter whereas the collision between the robot and human is relatively high frequency. By investigating the frequency characteristics of each part of the manipulator and possible disturbances, we proposes a set of band-width filters using intrinsic data and detects the human-robot collision while the robot is in motion for performing some tasks. We verify the effectiveness of the method using two different types of robotic manipulators; electric motor driven manipulator and manipulator with hydraulic rotary actuators. In both cases, experimental data show that the robot-human collision can be detected reliably and faster than simply using a joint torque sensor while the manipulator load is not constant.

Enhanced collision detection method using frequency boundary of dynamic model

This paper presents development and experimental evaluation of a collision detection method for robot manipulators sharing workspace with humans. For the safety of cooperative workspace, fast and reliable collision detection is important. In the same time, collision distinction is necessary to prevent false alarm for the performance of workspace. One of the main reasons of false alarm in collision detection is modeling error caused by work pieces or tools. The collision detection algorithm using band pass filter (Band designed Disturbance Observer, BdDOB) is used to eliminate this problem. We discovered the tendency of the manipulator dynamic models maximum frequency boundary during operations. This tendency is used to design the new BdDOB which contains changing frequency window. Thanks to the dynamics-considered filter band, the collision is successfully distinguished from other false alarm-candidates. In the result, the BdDOB shows not only fast but also robust collision detection performance under the condition of faulty sensor and uncertain model data. The experimental result from collision between a 7 dof serial manipulator and a car crash test dummy is reported.

Lane Marking Detection of Mobile Robot with Single Laser Rangefinder

Lane marking detection is one of important issues in the field of autonomous mobile robot. Especially, in urban environment, like pavement roads of downtown or tour tracks of Science Park, which have continuous patterns on the surface of the road, the lane marking detection becomes more important ability. Although there were many researches about lane detection and lane tracing, many of them used vision sensors mainly to detect lane marking. In this paper, we obtain 2 dimensional library data of `Intensity` and `Distance` using one laser rangefinder only. We design a simple classifier and filtering algorithm for the lane detection which uses only one LRF (Laser Range Finder). Allowing extended usage of LRF, this research provides more functionality not only in range finding but also in lane detecting to mobile robots. This work will be technically helpful for robot developers to design more simple and efficient autonomous driving system using LRF.

Joint Torque Sensor Embedded in Harmonic Drive Using Order Tracking Method for Robotic Application

The main contribution of this work is the low-cost joint torque measurement method using harmonic drive characteristics. To measure the external torque, we use the strain measurement of the flexspline of the harmonic drive. The flexspline of the harmonic drive is deformed by the external torque, as well as by the velocity reducing operation, which creates torque ripple on the strain signal from the flexspline. As the frequency characteristics of such torque ripple are strongly related to the joint speed, an order tracking technique is used to compensate the torque ripple. The resampling-based order tracking data acquisition is used to rearrange the torque ripple signal into the order signal, which has independent frequency characteristic to the joint speed. Then, a simple notch filter easily cancels the torque ripple from the rearranged signal. Our algorithm switches the signal source between the order tracking and the conventional time sampling to overcome the zero velocity condition. To overcome the lack of order information at low velocity, we propose a back propagation method using trained order sampling data. The developed joint torque sensing method is applied to the harmonic drive reducer of a robot manipulators joint and is experimentally evaluated for the torque measurement during the motion of 5th order polynomial trajectory following.

Torque Ripple Compensation Method for Joint Torque Sensor Embedded in Harmonic Drive Using Order Analysis

The main contribution of this work is an algorithmic compensation of joint torque ripple signals in harmonic drive for joint torque sensing from measuring the deformation of flex spline in the harmonic drive. When an external force is applied to a robot, the flex spline of harmonic drive undergoes an elastic deformation. However, such a deformation is not easy to measure by strain gauges on the flex spline because whose signals are corrupted by a periodic vibration which is “torque ripple” caused by reducing operation of the harmonic drive. We perform order analysis and design a filter to eliminate the torque ripple without using many strain gauges. As the order sampling normalizes the frequency characteristic, our method outperforms the filters with fixed cutoff frequency with manageable computational expense. Moreover, the effect of misalignment of the strain gauges is minimum unlike other methods. The experimental evaluation is provided using a robotic manipulator which has the proposed harmonic drive joint torque sensor.

Autonomous robotic street sweeping: Initial attempt for curbside sweeping

Autonomous street sweeping is an interesting robotic application because such street cleaning requires tedious and exhaustive human efforts of manual sweeping with brooms, blowing tools, or driving sweeping vehicles within a large area or along a long extension of roads on a regular basis. Toward autonomous street cleaning, we present our recent development of an autonomous robotic street sweeper. The developed system is aimed at autonomous cleaning with low cost hardware. In this work, a commercialized manual brush sweeper is robotized and equipped with two fish-eye cameras for environmental sensing. As an initial attempt of autonomous street sweeping, we show the cleaning performance of the robot along the curbside based on curbside recognition and tracking with top-view images.

SKKU Hand Arm System: Hardware and Control Scheme

In this work, we introduce the SKKU Hand Arm System I (SKKU-HAS-I) focusing on the hardware design and control scheme of the arm and hand systems. For the robot arm system, a driving module unit is designed and the workspace analysis is performed for the arm. A Virtual Spring Damper based controller is applied to the arm system for the task of the control. The design of the robot hand is based on mimicking the human hand and we perform an optimization process for three different design measures, workspace intersection volumes, manipulability, and opposing angles. The developed robot hand is equipped with various sensors for the contact information. Experimental results are provided for the evaluation of the developed robot hand.