Jialong Liu(刘家隆)

Jialong Liu(刘家隆)

Stay hungry, stay foolish

Portfolio

Research work on robot reinforcement learning

Published:

After completing my undergraduate studies,I joined Professor Chen Changhao’s PEAK_Lab at the HongKong University of Science and Technology (Guangzhou) as a research assistant,where I am responsible for research on leggedrobots.Currently,I focus on the application of deep reinforcement learning in robot control and the whole-body control of robots,dedicated to enabling robots to accomplish a series of complex tasks through reinforcement learning.

Development of dexterous hands

Published:

In the second semester of my senior year, I was fortunate to participate in the initial work of a startup company collaborating with laboratories from Harbin Institute of Technology, Shenzhen (HIT Shenzhen) and University of Science and Technology of China (USTC). As a robotic software engineer there, I specialized in ROS2 development and driver configuration (RS485/CAN) for a 22-DOF dexterous hand, along with the integration of the UR16e robotic arm. I led the development of ROS2 Jazzy/Humble versions, which included: - Motion control systems and sensor data processing pipelines - Communication module architecture for multi-device coordination

Emergency Response via Spherical - Ducted UAV Integration

Published:

In the last year of my undergraduate degree, I participated in the task of Project Design IV of the Guangdong-Hong Kong Robotics Academy, responsible for a flight control part that combines a spherical robot and a drone. This robot can shuttle through complex terrain and avoid obstacles by flying. It can not only conduct drone patrols in terrain such as forests, but also use the rolling characteristics of spherical robots to cross complex terrain areas.

Robocon competition experience

Published:

In the two-year ROBOCON competition,I participated in the competition in two different capacities: team leader and technical advisor.During the 23rd season,I was involved in the development of control codes for rabbit and elephant robots,as well as the development of the ROS segment.In the 24th season,I was mainly responsible for the development of decision-making algorithms for autonomous robots.
By the way,our team won the national first, second and third prizes. It is particularly worth mentioning that we achieved the 2nd place in the national preliminary competition in 2023!

Robot Basic Competition:2D radar algorithm design and control code development

Published:

This is the selection competition of the Automation Academy for entering the Robocon team of the Robotics Academy. We use the traditional three-wheel omnidirectional wheel as the chassis, plus two U8 LITE as the friction belt motor. In the process, 2D radar is used to detect the cylinder, and finally, the radar automatically detects the cylinder alignment and launch score.

A sports and health analyzer based on CNN-LSTM

Published:

This project addresses the growing need for accessible foot health monitoring and early disease detection. Inspired by national policies promoting healthcare innovation, it leverages lightweight sensors and AI algorithms to identify foot deformities (e.g., flat feet) and systemic risks (e.g., Parkinson’s) through gait analysis. Unlike bulky clinical tools or basic wearables, our solution offers real-time insights, personalized advice, and health alerts, aiming to empower users with proactive care.You can click here to view our implementation effect.

Manipulator control,sensor collection,model building for smart med - arms

Published:

This is a laboratory project at Guangdong University of Technology. The purpose is to touch the mass through the robotic arm and judge the size of the mass. In the experiment, we used the Jaka robotic arm combined with a six-dimensional force sensor to slide the target object (pork nugget) at different speeds, angles, and depths. Finally, we combined the LSTM algorithm to perform a long-term sequence analysis of these data. Similarly, a feedback control can also be performed on the robotic arm based on these data.