wild-galaxy-9169
Edited
• 3 Credit Hours
Key adjectives used by students — color intensity reflects sentiment
wild-galaxy-9169
Edited
timely-ferret-8552
Edited
patient-parrot-8453
Edited
fierce-orca-6183
Edited
mellow-walrus-6595
Edited
happy-rocket-5528
Edited
agile-salmon-0029
Edited
serene-giraffe-6567
Edited
friendly-viper-6211
Edited
gentle-puma-9412
Edited
Took this course in Fall 2025. Overall this is the best course so far I've taken at Gatech (I've completed 5 courses). I end up with an A, not even trying too hard cuz for both midterm and final term I only use one chance of two since I was too busy at work. I would say this course is Organized, Supportive, Rich in Content and having Manageable Workload.
My background: No previous CS background, only took a few fundamentals programming class, SW development class and ML class during my master degree. Which means I don't have systematical learning experience as well as strong skills in math. But yes, I'm quite familiar with python, have basic knowledge of matrix operation, limited knowledge of probability. Those things are enough to support me to understand the lecture and keep learning during the semester.
Class Content: Well structured. I won't say it is very updated to the trend but they are very useful and the lesson is well polished. You will learn basic algorithm of robotic (self-driving) which is widely applied to different areas. I took some note during the lecture and by reviewing the note, it's easy to grab the main knowledge quickly and apply them in the hw. Personally, I think it is an interesting class, neither too hard nor too easy.
Pset and Project: Small Pset is pretty easy and will only take 1-2 hour as long as you watched all the course video. And there are videos going through the problem set which you are allowed to reference in your code. They are also very helpful as most of the content will be related to the project and exams. Project will take relatively longer time but they also allowed you longer time on them. Most of the project took me around 10 hours, some of them are shorter. But not all 10 hours are coding, you may spend time watch the office hour recording where the TA will walk through the long project pdf and give you a few useful hints. After you take time fully digest what is the problem about, the coding time could be really short. And then, another time-consuming thing will be tuning the hyper-parameter. This could different you from get 85/100 or 100/100. If you are lucky, or you chose not having those 15points, you may use way less than 10 hours in each project (maybe only 6 hours). However the final 1-2 points sometimes really took time. It depends on your decision.
Exams: Instructor will give you some helpful tips on review. I would say as long as you watch the lesson video, did the project on your own, went through the knowledge mentioned by the instructor, it should be easy to get a 80/100 in both exam. Nothing far beyond the course and project. Both exams have 2 chances, I only took once as I was busy but still the grade is quite enough for me to get an A.
Instructor & TA: Very supportive. Most of the time I won't post anything on edX. But in this course, I've taken the discussion on edx as a main learning and Q&A resource as the stuff team will carefully answer any question especially regards project. Their response speed is also very quick. And most of the questions I have during the project, I can find the answer in the discussion before I even asked.
Overall, really recommend this class.
I took this course Fall 2025. I got an A (90~%). BG: BS in CS. 2 years SWE experience. This was my first class in the program. The lectures were okay. The instructor does NOT go into much detail. He tells you what needs to be done and lets you implement it in python, but there isn't any in depth theory or the algorithms which is a bit disappointing. HW: You're given the answers and can use them as is, which helps a lot. Projects: I found the projects to be the most challenging part. They were very much doable tho. Ed discussions and code reviews (I had 2) helped more than anything for projects, but make sure you read them frequently!! Even though some of the projects were hard, I enjoyed them, especially the particle filter one. Another thing, you can re-submit as many times as you'd like on Grade Scope. Exams: Pretty straightforward. Study the basic theory behind the concepts and you're good. There's no coding. Is it good as your first course? Yes, but you need to know python. Is it easy? Not entirely, some parts are hard.
If you take anything from this review let it be this: It is imperative that you start strong. Finish as many lectures and HW assignments as you can. You have a month for each of the first two projects (as far as I remember). Start them early. It only gets harder the more you advance in the course. If you want it to be easy, start early!!
Finished with an A.
Workload is not too crazy, exams are quite fair and they test closely to what is asked in the project and mirror the lectures (which are good). Projects are very fun! Take time to understand the lectures and then you will do well in the exam and the projects.
Overall, one of the better courses and I strongly recommend it.
4/5 because some lectures have errors, where you start doubting yourself and then you see at the bottom: "Actually, the lecturer did an error here and actually XXXX". It feels that rerecording some lessons and fix those errors would not take too much time.
Also, it was not updated in a while; I am sure that there are now other methods. Also, we talked a lot of about sensor fusing, which is central to many applications but we did not have a project or a lecture about it, which went a bit more in-depth.
Took this class in Fall 2025. This class had the best projects in the 4 classes I've taken so far. This is because they have a good balance of what they provide vs what they leave to you to figure out, plus it's very satisfying to finally succeed. There are plenty of resources between the lectures, tutorials, Ed posts, etc. to do well. However, this doesn't mean it's easy. There were a few times where I was really struggling trying to debug my project code (namely kalman filter, SLAM, and particle filter). Another thing I'll say is that this course is somewhat front-loaded, since the Problem Sets get finished about half-way through and there aren't many lectures after that. The only downside of this class is that the lectures can get pretty confusing at parts, but the staff does a good job at helping out when that happens.
Overall, RAIT is a very well organized course. As someone who's juggling a full time software engineer role and the OMSCS, this course helped me ease into the program. I wouldn't call this course a cakewalk, but extremely doable provided you put in effort.
======= Projects ======
Overall, the projects were super fun and a great way to apply the lecture material. Yes while the lecture material doesn't seem the most updated, I used these materials and the provided problem sets (the TAs provide the problem set solutions already) to get a good starting point for how to think of each of the projects above.
The exams also felt super fair. As long as you did the problem set diligently and understood how to implement the projects, you should be all good for the exam.
===== Course Staff ==== Honestly the TAs and Professor Summet made this course an awesome learning experience. In particular, John Liu's recorded sessions for each project were very energetic and gave really good hints for your project. Professor Summet is also very involved in the course and we see him in the Q&A office hours session. I recommend going to OH to get to know the course staff
====== Tips =====
As John Liu says, "don't get cute". Make sure you apply this advice! Trust me, if you keep your project code simple, you have less to debug and you may score better on the autograder. Write good helper functions/helper classes to keep your own code readable. Clear train of thought means clearer debugging for you
I found Kalman filters, Particle filters, and Warehouse search projects to be the simpler projects. These 3 projects are very doable provided you follow the course material closely (follow the videos in Canvas) and adapt the already provided problem set solution to the project. Tuning is the most annoying part, but if you go to the TA sessions, the TAs give you hints on "what ballpark you should be at for your constants"
Course was well structured and the syllabus was rational - you learn about robot localisation, path finding, PID, and SLAM (alongside other smaller topics). The projects were effective in forcing students to actually understand and implement these concepts, although the difficulty gap between the lectures (high level) and starting on the project can be quite big.
Exams were closed book, proctored MCQs which were challenging but fair. Other than these 2 major components, there's almost no busywork unlike other courses which was refreshing too.
Overall, good course to take during summer that won't kill you, but also challenging enough to feel like you did learn something useful by the end of it.
I took this course over the summer so I believe the pace of assignments is a little bit faster than in a normal semester. I have a background in CS, this was my 5th course in OMSCS and I ended up with an A. Overall, this course is very manageable and the autograder is fairly lenient for the projects, which is nice because it can get pretty frustrating tuning your solutions to pass all of the test cases. I felt that the projects in this course were satisfying and I do feel like my Python skills improved in the end. I feel like I gained a good introduction to some robotics concepts and was genuinely interested in the course content throughout. Additionally, the TAs were really active on the Ed forum. My complaints about the course are that the lectures have a lot of typos, which made them a little bit frustrating to watch at times, and as I already mentioned, the projects can be time consuming when you have a working solution and it boils down to tuning. I personally ended up just calling it good enough for a few of the projects and settling with a mid-range A instead of spending many more hours for the 100. For a few of the concepts, I also had to spend some extra time with additional resources they provided in the course and some external resources as well to really understand them. I found the exams to be relatively challenging but fair. You get two chances to take them, but the questions are different. And notably, I and many others found the second time to be more challenging. Whether that was random or not I don't know, but just keep in mind the second time could be harder.
My advice for people taking this course would be:
Overall, this course is worth taking in my opinion. I would consider this an easy A class if you just put in some effort. It was a fair pace for summer and I think if you have a lot of time to spend on school, it could definitely be paired with another course in Fall / Spring.
AI4R (now Robotic AI Techniques; RAIT) was a great survey course discussing some major approaches to robotic localization (where are you?) and control (what should you do?). There's enough math that a light stats and/or linear algebra refresher could be useful, as it will help you develop a better intuition for the concepts.
The midterm and exam are of middling difficulty, Honorlock proctored, and quite fair overall. The meat of the course is in the project and problem set work, where the problem sets will introduce practical code-based approaches to the concepts presented in lecture, and the projects provide a more substantial exposure and challenge involving the key concepts. The code framework is robust, with test cases and visualizers available to inspect your code's behavior. The overall course load, even for summer, is quite fair, though if you fall behind you're going to have a bad time. I personally advise trying to stay 1-1.5 weeks ahead, i.e. finish things a bit over a week before they're due, to ensure you have time to spend on the later, more difficult projects in the course.
It's a bit of a lighter course, but very interesting and fun. The TAs are extremely organized and helpful, and I would strongly recommend this course if you're considering it.
I got A. Yes this course as of now (25Summer) is purely a course to raise your GPA and likely can learn nothing (if you have modest AI basics and you likely have if you are CS grad as in 2025). Textbook is dead, also the Ed lectures. I never attended OH. HW and proj are easy for obvious reasons. Cant say this course means anything to robotics specialization. Also maybe OMSCS robotics specialization itself means nothing in the job market, to be honest. Maybe Seminars (reform) can make some difference...sigh
I have mixed feelings about this class. On one hand, I felt like it helped me a lot with my confidence in Python and my ability to code things from scratch (for the most part). On the other, the lectures felt outdated and full of small mistakes that made them frustrating to watch. The algorithms covered were pretty simple, but I really wouldn't even call them AI. The projects are the bulk of the class - you will either love them or hate them. There is a big disconnect from what is expected in the projects and what is shown in the quizzes/lectures. The projects are indeed satisfying, but the process to get there can be very frustrating. I strongly recommend anyone taking this class to go through the Ed Discussion forum, as you will find a lot of answers there - mostly from other students. I never once got a useful answer from a TA. This seems intentional as they don't want to "give away answers". However, some of their responses could come across as rude or condescending. I once saw someone ask a question along these lines: "how can we align the robot in this direction?". The TA's response was something like "using simple 10th grade math". I do recommend this class, but expect to spend a lot of time on the projects. The time given for each project is generous, and you can get ahead if you want to. My recommendation is to start early and distribute the workload so that you don't rip all your hair out. For each project, I had to spend 2-3hrs just trying to figure out how to even get started and write my first line of code.