Welcome to State Estimation and Localization for Self-Driving Cars, the second course in University of Toronto’s Self-Driving Cars Specialization. We recommend you take the first course in the Specialization prior to taking this course.
This course will introduce you to the different sensors and how we can use them for state estimation and localization in a self-driving car. By the end of this course, you will be able to: - Understand the key methods for parameter and state estimation used for autonomous driving, such as the method of least-squares - Develop a model for typical vehicle localization sensors, including GPS and IMUs - Apply extended and unscented Kalman Filters to a vehicle state estimation problem - Understand LIDAR scan matching and the Iterative Closest Point algorithm - Apply these tools to fuse multiple sensor streams into a single state estimate for a self-driving car For the final project in this course, you will implement the Error-State Extended Kalman Filter (ES-EKF) to localize a vehicle using data from the CARLA simulator. This is an advanced course, intended for learners with a background in mechanical engineering, computer and electrical engineering, or robotics. To succeed in this course, you should have programming experience in Python 3.0, familiarity with Linear Algebra (matrices, vectors, matrix multiplication, rank, Eigenvalues and vectors and inverses), Statistics (Gaussian probability distributions), Calculus and Physics (forces, moments, inertia, Newton's Laws).
HK
The course gave a clear and an in-depth knowledge on Kalman filters and Localisation using those filters. The assignments were pretty tough but solving them was fun.
MP
A lot of fun! I learnt a lot and feel that due to the well designed assignments I really got to the bottom of it...
MI
Very interesting course if you want to learn about the different filters used in self driving cars for sensor fusion
RL
It provides a hand-on experience in implementing part of the localization process...interesting stuff!! Kind of time-consuming so be prepared.
EJ
I have learned KF in the past. First time learning EKF. I liked the rigor in this course! Felt like a legitimate university lesson.
TM
This is an eye-opening course on how to utilize statistical analysis for engineering applications, and in particular, autonomous systems, as such it is very useful and captivating course!!!
H
A well-taught course by Prof. Jonathan Kelly.I accumulated huge amount of knowledge after undergoing his teachings.The supplementary readings proved to be of great help to ace the final project.
JG
Very challenging, nevertheless excelent for learning automation concepts, python programming, sensor fusion, probability & statistics
SI
Thank you for this absolutely fantastic course. Kalman filters and state estimation in general is a concept that I've tried to understand for a long time, and I'm glad to have finally understood it!
AQ
One of the most exciting courses ever had in terms of learning and understanding. Kalman filter is a fascinating concept with infinite applications in real life on daily basis.
FI
Challenging course, specially the assignments. The extra literature resources are great. The explanations and examples on the videos could improve. Step by step Hands On examples would fit great
ZR
Video lectures arer great. Programming assignments are also well designed. I just hoped more info of how input data for the last assignment was acquired.
Showing: 20 of 136
There is no support for this class
The forums are almost useless and no teacher or staff ever answers anything on them
The lectures are pure fluff and hand-waving, no meat and no details
The projects are extremely difficult and there is no lessons to cover material needed for the projects
Would not recommend unless you want to basically learn on your own
Too much work BTW I did get 100%.
Review :
Mentor Help: 0/5
Course Content: 4/5
Course Explanation: 4/5
Course Challenging: 4/5
Exercises : 3/5
Things which can be improved: There should be a programming exercise for each module especially for modules like ICP. There should be more mentor support as everything can't be understood by videos. There is/was an expectation of doing the final project in CARLA online but it was offline and also the ICP was pre-implemented. But overall for starters it is a very good course for state estimation to support and I strongly suggest to complete it if you aspire to be a self - driving car engineer.
Very dry lectures!
Quiz automated grader buggy and not working at times. Example: not well defined python environment for the quiz in module 4. A grader expects a certain format that you have to guess. But to guess you need to submit the quiz in order to see if you satisfied the grader. So you can do that 5 times every our. A lot of time spent on satisfying the grader format that learning material.
The reason I am realty trying to stay in the class is because I am very interested in the subject but the execution of this class is a disaster!
One of the most exciting courses ever had in terms of learning and understanding. Kalman filter is a fascinating concept with infinite applications in real life on daily basis.
There are many interesting topics. Without the help and suggested readings from this course, I wouldn't be able to finish by myself. Also, the final project is very enlightening.
one of best experiences. But the course requires a steep learning curve. The discussion forums are really helpful
Very useful!
Great experience!
Congratulation all the people involved in this course!
The course gave a clear and an in-depth knowledge on Kalman filters and Localisation using those filters. The assignments were pretty tough but solving them was fun.
It provides a hand-on experience in implementing part of the localization process...interesting stuff!! Kind of time-consuming so be prepared.
The course content is good, the instructors are good, and the projects are good. But I hate the quizzes and notebooks throughout the course that don't provide better guidance or step-by-step solution checking. It would be much better overall if quizzes and notebooks in the courses either provided step-by-step solution checking or provided the solutions so students could check their work along the way. I’d much prefer the notebooks provide the solutions or most of the solutions and have a difficult final project for each course where there were no solutions given. I’d learn much more through the course and have confidence while completing the final projects, and have a sense of accomplishment that I applied what I learned. I’m so frustrated with this that I’m cancelling my subscription for now.
I was impressed about the different methods available to do state estimation. The content was well presented (all slides shown are available as a PDF download) although in a quite compressed fashion. As in course 1 I would have preferred much longer videos so that more details of the different models could have been highlighted. Personally I was amazed about concepts like the Quarternion that I have never heart about before. A great plus from my perspective is that - like in course 1 - every lesson has a list of further articles to read - and in order to really comprehend the stuff presented I recommend in doing a deep-dive into these articles. Personally I found the coding assignments really demanding and as a side note I would have appreciated a little bit more presence of the teaching stuff to clarify. Currently the impression is that besides a monthly post in the discussion forum the teaching stuff is not visible - which is really sad as I think this whole specialization to be prime content. Unfortunately the locked video that will be shown to you when having completed the assignment is only a white screen and you are not able to follow the explanations the professor is providing. I would really appreciate if the invisible slides would be available for download but this is not the case. All in all I am a little bit mixed about the course as for example particle filters are just mentioned in one video but not explained as all the various types of Kalman filters. Still I give this course a 5-star ranking as it provides a good starting point for those trying to dig deeper into SLAM.
This is not an easy course. If you expect to understand just by watching videos, you are horribly wrong. Definitely read the suggested reading materials and derive things on paper as you go, at least that's how I learned. What I like about this course is that it provides me a relatively comprehensive introduction to sensor fusion (except for camera data, which will be covered in another course in the series). I wish there are more practices with radar data. In fact I had to decide to take Udacity Sensor Fusion courses mainly for that reason because radar will just become more and more important for true L4 and L5 in my humble opinion.
Challenging but very fun. Not for beginners, you certainly need to know your math and be good enough a coding. Very recommended introduction to state estimation.
Very interesting course if you want to learn about the different filters used in self driving cars for sensor fusion
Excellent course! The best course available of this topic
a very good course about sensor fusion ans localization
Very helpful!
great
I finished the first course in this specialization and now I'm in the second course. This is mostly a marketed course which doesn't have much academic value. They just have emphasized on beautiful slides and graphics. The content is shallow and doesn't go into details. If you are looking for a course that you learn the concepts well, this course is not for you.
too much high level, no in depth treatments of the topics. There are free course avaiable online which are more in depth than this (paid) course