Welcome to my technical portfolio! I'm Tri Bien , Robotics Lab Engineer at Vietnamese German University. I have an extensive background in mechatronics, software robotics, and robotic control. My research interest is the intersection of Computer Vision, Machine Learning and Optimization Motion Planning techniques for solving dynamic robot problems such as dynamic grasping or handover tasks, as well as developing autonomous intelligent robots based on ROS (Robot Operating System).
I graduated in 2017 from Karlsruhe University of Applied Sciences with a Master's in Mechatronics System and Sensor Technology. With a strong passion for the robotics field, I have worked on researching and doing several robotics projects, which are displayed here.
The research focuses on designing an Optimization algorithm for reactive motion controller manipulators which can avoid static and dynamic obstacles while moving to the desired end-effector pose. This work is based on a paper: "NEO: A Novel Expeditious Optimisation Algorithm for Reactive Motion Control of Manipulators" and the book "Learning for Adaptive and Reactive Robot Control". Video
The research focuses on designing a system for position-based visual servoing (PBVS) with dual 6-Degree of Freedom (DoF) UR10e robots. The main robot uses computer vision integrated position-based servoing algorithm for tracking the ARTag trajectories from another robot by using Realsense D435 camera. The result of this algorithm is joint velocity control on the real UR10e robot, which helps the main robot simultaneously track the ARTag trajectories in real time. Video , PrePrint
This research aims to develop a lidar-based vehicle detection system in a complicated roadway area through an embedded system. In order to achieve this, a low-power, high-performance embedded GPU (Nvidia JetsonTX2) with support for deep neural networks running concurrently and the application of an unsupervised learning DBSCAN algorithm for point cloud recognition has been chosen. Video, Preprint.
This study uses a Vicon system, a high-speed accurate marker-based motion capture system aiming to develop a wireless low-latency gesture control method that requires no hardware development and no on-board power source. A novel control is designed to track a rigid-body object’s motion in the work- space and translate it to robot control signals. The system has five main compo- nents: a host PC with a Vicon software named Tracker, a set of eight Vicon Van- tage V8 infrared cameras, a client PC that receives Tracker’s broadcasted data over the Ethernet, a robot model named youBot provided by KUKA, and a rigid- body object that will act as the gesture controller. Video, Preprint.
MiniRos is a small autonomous UGV-Unmanned ground vehicle developed for education and research purposes. In this paper, we develop MiniRos with a kinematic skid-steering, high torque and compact structure drive system that can take research to the outdoor environment. MiniRos is run on a Robot Operating System (ROS), which has been developed to be compatible with standard robotics programming environments. The robot can build a 2D map with various SLAM algorithms such as Karto SLAM, Gmapping, Hector SLAM, and navigation in the built map. Video, Preprint, Code.
The study shows the new platform of the V-frame octocopter kinematics analysis, designed on the CAD software, with some important mechanical parts using FEM analysis to find the highest stress and displacement under high load applied, the result of all connecting the joints 3D printing part is completely safe. Mechanical parts were manufactured by using 3D printing technology and CNC milling. Moreover, the study has shown V-frame octocopter simulation based on Simulink using the second method Ziegler- Nichols to find suitable parameters of the PID controller for roll and pitch angle. Using the block simulation is good for implementing and fast checking the new algorithm when building the new platform of the robot. Video, Preprint.
In this study, a new optimal control based on saturation function for the upper exoskeleton is presented. The saturation model is an advanced model including signum function property, which is symbolized by the sliding mode model. Input control of the system is then designed following the saturation model. The gain of the input control is obtained following adaptation law related to the dynamic parameter of the system. After formulating, the control is simulated for evaluation. The random movement of the upper exoskeleton is applied in the simulation, and the tracking performance is used for the proposed controller. The simulation results show that the proposed control is good performance and can be applied in the real system.
The Upper body Humanoid robot includes 14DOF (Head 6-DOF and 2x Hand 4-DOF), which integrated 3D depth camera. This model can manufacture by 3D printer or laser cutter method. This robot designed for research and educational purposes. 3D File
This autonomous service robot can be used in restaurants, hotels, hospitals, and industry. The robot is implemented using ROS Middleware, 2D Lidar, an Embedded computer, and various sensors. The robot can perform SLAM and Autonomous Navigation. Video
This design includes the charger docking and locking mechanism for the electric motorbike in "Bosch Green Challenge project", I got awarded "Certification of Innovation Activities and Development" for this design. 3D File
The robot has 23-DOF and each leg has been designed with 3 DOF. Thus, a total of 18 motors a required for walking and operating an Ant-like robot. I used a servomotor actuator and plastic frame for the leg mechanism, the robot was used for STEM teaching. 3D File
Humanoid robot with 19 DOF for STEM teaching, This model can manufacture by 3D printer or laser cutter method. 3D File , Video
