Shuhao Qi | University of Melbourne

Research Interests

Robot Control and Decision Making: Optimal Control, Reinforcement Learning
Robot Perception and Artificial Intelligence: Signal Processing, Patern Recognition, Deep Learning

[Oct 2021 - Present] Shanghai AI Lab
Intern Student, waiting for Australian Visa
- Project: Decision Making for Autonomous Driving
[May 2020 - Jul 2021] Southern University of Science and Technology
Research Assistant in CLEAR Lab
Principal Investigator: Prof. Wei Zhang
- Project: Motion Control of Legged Robot
[Sep 2017 - Jan 2020] Beihang University
M.Eng. in Control Science and Engineering
Advisor: Prof. Weihai Chen
- Thesis: sEMG-based Extraction of Motion Intention and Assist-As-Needed Control of Rehabilitation Robot
[Sep 2013 - Jul 2017] University of Electronic Science and Technology of China
B.Eng. in Mechanical and Automation Engineering
- GPA: 3.78/4.00, Rank: 12/280
- Thesis: Laser-SLAM based on Iterative Closest Point and Graph Optimization

Perceptive autonomous stair climbing for quadrupedal robots
Shuhao Qi*, Wenchun Lin*, Zejun Hong, Hua Chen, Wei Zhang
The IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2021. [Video]
sEMG-based recognition of composite motionwith convolutional neural network
Shuhao Qi, Xingming Wu, Weihai Chen, Jianbin Zhang, Jianhua Wang
Sensors and Actuators, A: Physical (Q1, IF=2.904). [PDF]
Recognition of composite motions based on semg via deep learning
Shuhao Qi, Xingming Wu, Jianhua Wang, Jianbin Zhang
S 14th IEEE Conference on Industrial Electronics and Applications (ICIEA 2019). [Slides]
Damping vibration analysis of adual-axis precision force sensor based on passive eddy current
Xiantao Sun, Wenjie Chen, Weihai Chen, Shuhao Qi, Jun Jiang, Cungang Hu, Jun Tao
Journal of Physics D: Applied Physics.
Design and analysis of a large-rangeprecision micromanipulator
Xiantao Sun, Wenjie Chen, Weihai Chen, Shuhao Qi, Wang Li, Cungang Hu, Jun Tao
Smart Materials and Structures.

Decision Making for Autonomous Vehicle(Reinforcemnet Learning)
Oct. 2021– Present
We are trying to develop a method to learns high-level driving decisions by imitating low-level control behaviors.

Stairway Climbing of Quadruped Robot(Optimal Control)
Jun. 2020–Nov. 2020
We develop a simple yet effective framework tailored for autonomous quadrupedal stair climbing. By exploiting the prior knowledge about the staircase structure, the proposed strategy first extracts the geometric information about the staircase from the depth measurements. Then, the velocity and foothold references during stair climbing are generated via simple optimization algorithms based on the staircase information. Given these references, we use the model predictive control method to generate joint torques to complete the whole stair climbing task. Simulations in MuJoCo and real-world experiments on the real Aliengo robot demonstrate successful autonomous climbing of various staircases with different geometries.

sEMG-based Active Control of Rehabilitation Robot (Force Control)
Jan. 2019–Jan. 2020
In order to make people participate in rehabilitation exercise actively, I proposed a scenario which can adjust the joint torque of the upper-limb rehabilitation robot according to the amplitude of sEMG signals. [Video]
sEMG-based Motion Recognition (Deep Learning)
Jun. 2018–Jan. 2019
Compared with currently intricate and limited methods, the method I proposed is much simple, effective and general which processed sEMG signals like the way of processing images via Convolutional Neural Network. Results of experiments show that this method can help people effectively recognize general composite motion, such as handwriting motions and sign language motions.[Video]

Laser-SLAM based on ICP and Graph Optimization (SLAM)
Oct. 2016–Jun. 2017
Motivated by Cartography of Google, I established a scan-to-map Iterative Closest Point (ICP) algorithm based on triangulation and SVD algorithm. Additionally, least square method was used to optimize graph after close-loop detection. Simulation shows that this method achieved excellent performance.
RoboMasters Competition (Embedded System)
Mar. 2015–Jul. 2015
Representing UESTC to participate in this national competition organized by DJI, and taking charge of “Gunner” robot, which can accurately hit the target from a distance with the help of visual positioning.

Automatic Packaging Machine (Embedded System)
Feb. 2016–May 2016
I, as a member of robot team, represented my university to participate in this national competition organized by DJIand took charge of “Gunner” robot, which can accurately hit the target from a distance with the help of visualpositioning.[Video]