Classes
- MAE 435 Principles of Automatic Control (Undergraduate)
- MAE 789 Advanced Robotics (Graduate: advanced course)
- MAE 589 Modern Robotics (Undergraduate/Graduate: basic course)
MAE 589 Advanced Robotics (Advanced)
This class will be offered in Spring semesters
Discription:
As robotics is being actively utilized and applied across numerous applications, Model-Based and Data-Driven methodologies have become widespread. Many robotic systems, including humanoid robots, quadruped robots, manipulators, mobile robots, and drones, execute their missions by employing specialized planning and control techniques or by learning policies based on their distinct characteristics. This course introduces fundamental principles such as Caculus of Variation, Pontryagin's Maximum Principle, Model-Based Optimal Control, Linear Quadratic Regulator, and Safety-Critical Control, including control barrier functions, alongside data-driven approaches including the Markov Decision Process (MDP), Reinforcement Learning, and Proximal Policy Optimization (PPO). Our specific objective is to interpret and apply these core control and learning methods effectively within the context of robotics.
Format:
This course will be structured around two weekly lectures that will cover advanced concepts and theory, including optimal control theory, planning and control techniques, and data-driven learning methods, with a particular focus on methodologies applicable to legged robots (including humanoid and animaloid robots) or manipulators. Students will be required to complete two homework assignments emphasizing theoretic foundations. The course assessment will also include two midterm exams to evaluate theoretical understanding, and a final project where students will select one of three given problems and solve it using simulation tools.
Prerequisites:
These classes are not mandatory but encouraged.
Linear Algebra, Linear Systems, or Similar Math Topics
Dynamics Class, e.g., MAE 208: Engineering Dynamics
Control Class, e.g., MAE 435: Principles of Automatic Control
Basic Robotics Class, e.g., MAE 589: Modern Robotics
Lecture Slides and Videos:
Lecture slides will be posted on the course website one hour before each lecture. For students enrolled in the course, recorded lecture videos will be posted after each lecture.
Timeline:
| Date | Lecture | Deadlines | References |
|---|---|---|---|
| Week 1 | Course introduction | ||
| Week 2 | Review of Linear Algebra and Linear Systems |
|
|
| Week 3 | Calculus of Variations 1 |
|
|
| Week 4 | Calculus of Variations 2 |
|
|
| Week 5 | Optimal Control 1 | Homework 1 |
|
| Week 6 | Optimal Control 2 (Manilulation Applilcation) | Midterm 1 |
|
| Week 7 | Safety-critical Control 1 |
|
|
| Week 8 | Safety-critical Control 2 (Safe Navigation Application) |
|
|
| Week 9 | Markov Decision Process (MDP) | Homework 2 |
|
| Week 10 | Reinforcement Learning Concept (Humanoid Application) |
|
|
| Week 11 | Q-Learning | Midterm 2 |
|
| Week 12 | Policy Gradient Methods |
|
|
| Week 13 | Poximal Policy Optimization |
|
Grading and Course Policies:
Attendance is strongly encouraged. However, if you need to miss class due to reasons such as attending conferences or research-related events, please email Dr. Lee in advance with your excuse. We will have one or two in-class quizzes, all based on class materials. Therefore, it is essential to complete and thoroughly review the provided materials and contents.
| Attendence(Quiz) | Homework | Midterm Exam 1 | Midterm Exam 2 | Final Project |
|---|---|---|---|---|
| 5 % | 10 % | 25 % | 30 % | 30 % |
Project requirements:
A detailed description of the projects will be posted and announced once the semester begins and the course is underway.
MAE 589 Modern Robotics (Basic)
This class will be offered in Fall semesters
Discription:
Robotic systems have increasingly integrated planning and control algorithms to operate effectively in the real world. Legged robots, such as quadrupedal robots and bipedal humanoid robots, are designed to navigate challenging terrains and perform a wide range of loco-manipulation tasks, including autonomous inspection, transportation, and more. This course covers the fundamental principles of planning and control for legged robots. Topics include kinematic and dynamic modeling, floating-base dynamics, contact dynamics, hierarchical control frameworks, whole-body control, locomotion control, and trajectory generation techniques. These concepts will be illustrated with practical examples from quadrupedal and bipedal robots. The course emphasizes practical and applied approaches to planning and control, focusing on quadrupedal and humanoid robots.
Format:
The course will include two weekly lectures, four homework assignments, and a final project. The lectures will cover fundamental concepts such as rigid-body dynamics, optimal control theory, and planning and control techniques for modern robotics, with a particular focus on methods applicable to legged robots, including humanoid and animaloid robots. The assignments will emphasize practical implementation, requiring students to do simulation of locomotion gaits and solve coding problems related to the covered topics. For the final project, students will define their own problem and develop at least one solution using simulation tools.
Prerequisites:
Dynamics Class, e.g., MAE 208: Engineering Dynamics
Control Class, e.g., MAE 435: Principles of Automatic Control
Lecture Slides and Videos:
Lecture slides will be posted on the course website one hour before each lecture. For students enrolled in the course, recorded lecture videos will be posted after each lecture.
Timeline:
| Date | Lecture | Deadlines | References |
|---|---|---|---|
| Week 1 | Course introduction | ||
| Week 2 | Terminology, Vector Space, Transformation |
|
|
| Week 3 | Screw Theory, Kinematics, Singular Value Decomposition |
|
|
| Week 4 | Adjoint Representation, Single Rigid-body Dynamics | Homework 1 |
|
| Week 5 | Multi-body Dynamics |
|
|
| Week 6 | Joint-space and Task-space Control | Midterm 1 |
|
| Week 7 | Interpolations, Motion Planning |
|
|
| Week 8 | Model-free Path Planning | Homework 2 |
|
| Week 9 | Data-driven Planning |
|
|
| Week 10 | Centroidal Dynamics and Contacts | Homework 3 |
|
| Week 11 | Optimization-based Control | Midterm 2 |
|
| Week 12-13 | Whole-body Control |
|
Grading and Course Policies:
Attendance is strongly encouraged. However, if you need to miss class due to reasons such as attending conferences or research-related events, please email Dr. Lee in advance with your excuse. We will have four in-class quizzes, all based on homework assignments and class materials. Therefore, it is essential to complete and thoroughly understand all homework and review the provided materials.
| Attendence(Quiz) | Homework | Midterm 1 | Midterm 2 | Project Milestone | Project Final Report |
|---|---|---|---|---|---|
| 10 % | 10 % | 25 % | 25 % | 10 % | 20 % |
Project requirements:
You need to choose a proper high-fidelity simulation platform such as one of the following platforms:
Mujoco:https://github.com/google-deepmind/mujoco
Gezebo:https://classic.gazebosim.org/
Bullet:https://pybullet.org/wordpress/
Issac Lab:https://github.com/isaac-sim/IsaacLab
Research-inspired Project Topics:
1. Quadruped Robots (Unitree Go2) : A project on quadruped robots could explore how four-legged robots move and maintain balance. Students might investigate how different gaits—such as walking, trotting, or bounding—affect stability and energy efficiency. The project could involve building simple simulation models or using an existing quadruped robot to test basic control strategies. This topic is ideal for learning about locomotion, sensor feedback, and how animals inspire robot movement.
2. Manipulators (Franka Research 3): A manipulator project could focus on using a robot arm to perform everyday tasks like picking up, moving, or sorting objects. Students can learn about joint control, coordinate transformations, and path planning. A basic version might involve controlling a 7-DOF arm using joystick or computer commands, while more advanced options could include integrating cameras to detect object locations or implementing feedback control to improve precision.
3. Humanoid Robots (Unitree G1) : Humanoid robot projects offer a chance to study human-like movement and interaction. A project could involve making a humanoid robot perform simple tasks such as waving, walking in place, or balancing while carrying an object. This topic introduces students to complex coordination between joints, balance control, and planning motions that resemble human behavior. Simulations can be used if real hardware isn’t available, making this topic accessible to a range of skill levels.
Details for Project
1. Group Creation : Students are encouraged to define their own project in consultation with the instructor. Project groups should consist of no more than three members, based on the chosen topic and its level of difficulty.
2. Project Survey : Students will choose a topic from the list above and decide whether to work individually or as part of a group.
3. Project Proposal : Students will define their own research problems. To demonstrate the significance and challenges of the proposed project, at least 5 relevant academic papers must be cited. (3 pages PPT)
4. Project Milestone : Students will present the progress of their projects, including both theoretical development and practical demonstrations. The presentation must clearly indicate the overall completeness of the project and outline any challenges that may prevent full completion. (3 pages PPT)
5. Project Presentation/Report : Students will deliver a final presentation covering the entire project—from problem definition to validation (10 pages PPT). They must submit a final report formatted according to the IEEE conference template, with 5 pages for content and 1 additional page for references (more than 15 papers)