Carnegie Mellon University: Robotics Courses Reference¶
Reviewed against official sources: October 18, 2026. This guide connects CMU robotics topics with exercises you can run in the roadmap. Course descriptions and public notes are useful references; university registration, equipment access and assessed credit are separate.
Begin with Robotics Foundations if frames, feedback or planning are new. You can complete its CPU exercises without buying a robot or enrolling at CMU.
1. Bachelor of Science in Robotics (BSR)¶
CMU's School of Computer Science began the BSR in Fall 2023. The program overview emphasizes mathematics, sensing, mechanisms, control and integrated robot projects. The 2026–27 B.S. catalog is the source for the 360-unit degree and its full requirements.
The robotics core includes:
| Course | Focus |
|---|---|
| 16-280 or 16-281 | Intelligent Robot Systems or General Robotics |
| 16-220 | Robot Building Practices |
| 16-299 | Introduction to Feedback Control Systems |
| 16-384 | Robot Kinematics and Dynamics |
| 16-385 | Computer Vision |
| 16-450 → 16-474 | Robotics Systems Engineering followed by Robotics Capstone |
16-281 is the current number for the former 16-311, effective Spring 2026; it is not a future replacement still to arrive. The catalog also includes programming/systems, calculus, linear algebra, probability, 16-211 robotics mathematics, ethics and electives. Its sample sequence assumes prior introductory programming and a semester of calculus, so it is not a starting plan for a complete beginner. Consult the catalog for the full schedule and approved probability options rather than treating a shortened course list as degree advice.
2. Robotics minor & additional major¶
The current minor catalog lists 15-112 as the eligibility prerequisite and strongly recommends 15-122. It combines an overview course, controls, mechanisms/manipulation, building practice and two electives. Its 2.5 QPA and double-counting rules apply to enrolled CMU students, not to completing this roadmap.
For the additional major, use that program's current curriculum and application links. Application windows and course substitutions can change; an undated January–February deadline copied from an older page is not reliable enrollment guidance.
3. Key robotics courses and accessible material¶
| Reference | Verified term or source | Useful scope and access boundary |
|---|---|---|
| 16-281 General Robotics | Spring 2026 page; legacy URL retains 16311 |
Public schedule, homework/lab descriptions and prerequisite knowledge; team construction labs need physical components |
| 16-299 Feedback Control | Spring 2026 | Public notes/code and some recordings; includes balancing-robot work and simulation |
| 16-385 Computer Vision | Spring 2026 | Public course information; signup uses a Canvas code, so not all resources are guaranteed public |
| 16-831 Introduction to Robot Learning | Fall 2026 | Imitation learning, RL, visual decision-making and sim-to-real; course logistics and access are separate from the public overview |
The 16-299 page describes a revision toward control, planning and learning while retaining the existing catalog title. Some recordings are explicitly missing or incomplete. Read its topic/date labels rather than assuming every linked note belongs to one fully recorded current course. Its hardware assignments are not required for the roadmap's smaller CPU control exercises. Spring 2026 source.
The 16-831 Fall 2026 logistics identify Piazza, Gradescope and a homework repository. A public course homepage alone does not establish access to graded submissions, solutions or teaching support. The catalog's older 16-831.github.io course URL returned 404 during this review; use the term-specific page above.
Other useful catalog courses are listed below. A catalog entry confirms a course description, not that a particular term has a complete public course package.
| Code | Course/topic |
|---|---|
| 16-180 | Concepts of Robotics |
| 16-211 | Foundational Mathematics of Robotics |
| 16-350 | Planning Techniques for Robotics |
| 16-362 | Mobile Robot Algorithms Laboratory; multirotor simulation |
| 16-663 | F1Tenth Autonomous Racing; vehicle hardware labs |
| 16-664 | Self-Driving Cars: Perception & Control |
| 16-735 | Ethics and Robotics |
| 16-884 | Deep Learning for Robotics; research discussion and project |
For 16-350, the catalog expects introductory robotics plus programming/data structures. For 16-362, it expects vectors, matrices, coordinate systems and proficiency in an object-oriented language. These are substantive foundations; “introduction” in a course title does not mean no prerequisites. The catalog describes ROS in 16-280 but does not specify a ROS 2 distribution, so use the roadmap's compatibility-tested ROS material for installation.
4. A connected self-study route¶
The order below is our learning recommendation, not a claim that these labs replace CMU courses or degree requirements. Mechanical construction can run alongside software study when equipment is available.
| Stage | CMU topic reference | Do this in the roadmap |
|---|---|---|
| Programming and mathematics | 15-122 preparation, 16-211 | Foundations: vectors, frames, joints and timing |
| Kinematics and feedback | 16-384, 16-299 | CPU robotics lab: predict an arm position and explain a delayed controller |
| Dynamics and constrained execution | 16-299, 16-384 | Dynamics/task lab: torque limits, LQR/MPC, collision and grasp-frame checks |
| Perception and estimation | 16-385, 16-299 | AI/vision reference plus timestamp/frame-aware integration |
| Planning and software integration | 16-350, 16-280/16-281 | Explain the foundation lab's A* path, then run the optional local ROS 2 lab on its supported environment |
| Robot learning | 16-831 | Train a small CPU policy, then study Deep RL |
A concrete readiness check: a camera's x-axis points along the robot base's y-axis. A camera-frame displacement of (10, 0) millimeters becomes base-frame (0, 0.01) meters. Translating the frame origin does not change a displacement, although it does change a point's coordinates. If that distinction is unclear, work through the frame lesson before applying a learned action tensor to a robot. If the math is clear but the command arrives late, the missing topic is timing and control, not a larger neural network.
After the CPU route, choose a physical integration only when you can explain the sensor frame, action units, observation age and fallback behavior. CAD practice, assembling a robot and operating a controller on real hardware remain separate skills from a successful simulation.