Volver a Robotics: Aerial Robotics

4.5

2,037 calificaciones

•

526 revisiones

How can we create agile micro aerial vehicles that are able to operate autonomously in cluttered indoor and outdoor environments? You will gain an introduction to the mechanics of flight and the design of quadrotor flying robots and will be able to develop dynamic models, derive controllers, and synthesize planners for operating in three dimensional environments. You will be exposed to the challenges of using noisy sensors for localization and maneuvering in complex, three-dimensional environments. Finally, you will gain insights through seeing real world examples of the possible applications and challenges for the rapidly-growing drone industry.
Mathematical prerequisites: Students taking this course are expected to have some familiarity with linear algebra, single variable calculus, and differential equations.
Programming prerequisites: Some experience programming with MATLAB or Octave is recommended (we will use MATLAB in this course.) MATLAB will require the use of a 64-bit computer....

Oct 23, 2017

The course is very good.\n\nThe classes are well taught and show general concepts. It is necessary to do research on the internet, to solve the assignments. This is not a bad thing in my point of view

Jun 09, 2018

I think this is very good course of aerial robotics research. Being a student of robotics, I feel that some of stuffs in this course needs a good background in control and mechanical engineering.

Filtrar por:

por yukti s

•Apr 21, 2016

Assignments for week 3 and 4 were very tough and a little guidance was provided on how to do the coding part. But since I'm interested in the course i'm re-enrolling to complete it this time.

por Thadeus R

•Mar 14, 2016

This course definitely requires prerequisite knowledge of calculus and matrix math. There is also very little in the way of practical application. The MatLab work is interesting but barely explained. The assignments were nearly impossible without help from other students. Understanding which equations that were shown in the lecture (not necessarily explained) were useful in completing the assignments was not apparent and there was no help from the TAs.

I don't think the assignments were irrelevant but without having a background in academic engineering the course needs to walk through similar exercises with explanations to increase understanding and ability to complete the course.

While I don't think this course was well presented and I can't say I truly learned anything I am glad I was exposed to the material and I will continue with the Specialization.

por John T

•Mar 12, 2016

A tough score, which I have mixed feelings about as there was good stuff in here too. The course material is interesting and moves at a robust pace and I do think they have made an effort despite the fact that much of it appears lifted from one PhD student's dissertation. Realistically I would not recommend this course if you don't have a STEM Bachelor's degree and you will likely find it painful if you have been away from your degree more than three years. None of this is bad, although the material would have benefitted by being spread out by perhaps two more weeks as realistically people who have been in the workplace a long time may need more time, and have less time per week with family commitments. Unfortunately there are a number of areas that need work:

1 - Inconsistencies and errors in the material. Certain unexplained suffixes in equations and worse, changes in the suffixes without indication that they changed or what they are. I don't feel that this was particularly widespread but it did result in some loss of confidence in the course and time being wasted "interpreting"

2 - Tests throughout the course that often provide the relatively limited feedback of "correct well done" or worse I'm sorry to say, the relatively useless feedback "sorry that is not correct", without ANY explanation of why it was correct or more importantly what an incorrect answer should have been and why. I can understand that this will hopefully drive students should do more research, but if they hit a wall, realistically they're going to keep iterating on the answers until you pass and learn nothing because of the time pressure to complete by the end of the week. I wonder if there is a better mechanism that can be used here

3 - TA Support - The lack of TA support coupled with some concern about a history of errors led the students to believe that there was an error in week 3. For 10 days students went back and forth debating which one of the two equations that were supposedly doing the same thing but with missing terms were correct. NOT ONCE did a TA wake up and step in. In the end one of the students flagged the video as "inappropriate" to wake the UPenn organization up. The TA then stepped in and said (I paraphrase) "oh, we just dropped those terms because they're not so important, but we didn't mention that..." If you are not going to adequately support the students, the material had better be bullet-proof and show some linear thinking

4 - The last exam. Keep in mind if you do this course, you had better be comfortable with calculus, linear algebra, vector math/mechanics and it would be helpful to have a head start in Matlab. That said, the last question in the last exam, was an order of magnitude more challenging than everything else set and almost felt like a "shake out" question. I passed the course and had a good understanding of the material, but I suspect that the folks that did, made it through that last question in multiple random fashions. The material itself is relatively academic and the trajectory topic was definitely so. Unfortunately the one example (jerk trajectory) provided appears to have left a lot of students feeling very unsupported based on feedback I saw, and would probably benefit from having an example more fully worked through. As for the final exam, it would be highly desirable for UPenn to provide insight into how they would have solved the last part of the last question as my concern is that there is a whole contingent of people who did this course who didn't come away with as good a grounding on trajectories as they may believe they have

por Jiaming S

•Feb 09, 2016

The material this course provided is sufficient for a beginner in robotics. But I have to say that the material the coursed presented is not well prepared. The lectures are not well organized. Some necessary prerequisite should be at least provided as reading materials. Quizs ask some questions that is not clearly mentioned in the lecture and some are with a little ambiguity.

por Antón R V

•Mar 15, 2016

This course successfully covers the mechanics and control topics of quadrotors. The course also points to some resources in order to extend your study on UAVs. However, there are some issues concerning the assignments which I didn't like at all:

1. Big gap in difficulty between assignments of first three weeks compared with the assignment of the fourth week.

2. The difficulty of the assignments of the first three weeks just relied on sweeping (somewhat 'smartly') a 1,2,3, or 6 dimensional parametric space. I guess that real UAVs engineers are not paid for this. In some assignment, I had to edited a file which I was not supposed to edit in order to pass it.

3. The last assignment was quite frustrating for the following reasons

3a. Really long running times to test if the parameters work fine. I would sacrifice the fancy real-time visualization to save some time. Apart from long running times, I guess that UAVs are not solving ODEs in real time on board, so I think is more interesting a more practical/real-time approach.

3b. Search in a 12-dimensional parametric space (some parameters are equal, but still you end up tunning 5,6 independent parameters, which influence each other). Come on, is this useful? I am sure that UAVs engineers don't spend their hours in this kind of crappy search. You could explain at least some existing heuristics when having so many control parameters to tune.

3c. Even though the drone trajectory seems to fit perfectly the planned trajectory, you might not pass the exercise, and you would get some clueless messages telling you the final position/velocity is not correct.

3d. The assignment guide is very unclear in some parts.

3e. A lot of typos in the formulas of the last part of the guide. I decided to completely ignore it and develop my own strategy, which turned out to be far simpler and easier to implement.

3f. The submission program tests the whole exercise regardless you passed some sections or not, which make the waiting times even longer.

por Glenn B

•Feb 29, 2016

My feeling is that the course creator(s) did not fully prepare and scope of the course materials (i.e., lectures, supporting artifacts, assignments) required for the available time alloted to each week's topic. No syllubus or suggested reading material was available until the course started, which leaves little time to decide on the value of the course or to acquire the reading materials. The lectures glossed over topics that where only minimally supported by brief supplementary lectures provided by the teaching assistants. At the start of the course the lectures and sparse supporting material were only available in video format; not very conducive for detailed review. The student population had to beg for electronic copies of the materials, which became available around the end of the second week of the four week session. The assignments say they require 3 hours, but I would venture to say that most students have spent way more than 3 hours on the assignments just in researching solutions let alone debugging supplied software along with their portion of the assignment. In summary, the topics were interesting however the organization and supporting materials were lacking making for an unsatisfying learning experience.

por Ataliba M

•Jan 02, 2017

Very poor and a complete lack of interaction from Mentors/Tutors. Learners are left behind shooting bullets at the sky hoping to hit a star that might fall over with some hints. The course materials is consistent and very well informed but, as mentioned mentors/tutors since course began never showed up to clarify some frequent issues with Matlab codes and input data.

Felt disappointed!!!

por Masahiro S

•Nov 08, 2016

I'm sorry, but I felt the week 4 assignment materials is not consistent to lesson videos or the lesson video does not explained enough.

I think that making students think does not mean making explanation indistinct.

por Ziming W

•Aug 08, 2016

Interesting course. But definitely not for beginners. With a designer background, I could never accomplish this course. I really hope Penn could offer course in such topic for beginners!

por thiago c

•Mar 16, 2018

Quizzes and assignments are not tied well to the lectures. The flow and sequence is not good. The subject is very interesting, but needs to be better organized. the last assignment requires an unreasonable amount of knowledge of Matlab. I code in other languages and I am not a big fan of Matlab which make it even harder.

por Steve A

•Nov 10, 2018

Course is very incomplete.

por Nick J

•Jan 21, 2016

Requires MATLAB. In my view, a university should not be promoting a proprietary product. There have been other Cousera courses which used Octave, the free-software implementation of the programming language implemented by MATLAB.

I have seen no good reason why Octave should not have been used for this course.

por Zachary H

•Feb 20, 2016

So close but yet so far ...

Course concepts are interesting and the programming assignments are fun but the presentation can be greatly improved.

Criticisms include:

(i) The course isn't self contained. Physical concepts like 'resultant moment', 'inertial and body-fixed frame' and 'torque' are used without definition. Mechanics is not listed as a prerequisite but it should be.

(ii) Lectures are very mathematical but proofs, intuition and good problem sets are all missing. Listening to a math lecture without doing challenging problems or deriving mathematical results to build intuition can be a waste of time. Check out John Cochrane's Asset Pricing 1 and 2 or Tim Roughgarden's Algo 1 and 2 for great examples of thoughtful problem sets and intuitive derivations.

(iii) Way too much powerpoint! Speed reading a static powerpoint slide overloaded with dense mathematical formulas without using pointers or animations to focus the students attention is a recipe for confusion and frustration. I find hand written derivations, even when the handwriting is a little sloppy, much easier to follow than a static page of formulas plus a sound track. Hand written derivations impose a natural pace and focal point to the content. Check out Gilbert Strang's Linear Algebra, Sebastian Thrun's Artificial Intelligence for Robotics and Andrew Ng's Machine Learning for examples of good derivations of mathematically sophisticated material.

(iv) The programming assignments while fun were somewhat ad hoc and disconnected from the lecture material, specifically, the main task of every single assignment was to hand tune a pd controller. No systematic approach was ever described for performing this task.

por James L

•Feb 21, 2016

This course was not a beginning level course as advertised.

Test cases for assignments were neither document nor otherwise explained within the course materials. Quizzes during the lecture were often unrelated to the material just presented or were given before the relevant material was taught.

The course creators obviously put a lot of time and effort into this course. So it is difficult for me to write such a critical review. However, they missed the mark with this one enough that I can't recommend the course.

Unfortunately I will not be completing this course only due to the fact that I have run out of time and patience to complete the last assignment.

por Olena D

•Feb 29, 2016

I had very high hopes for this course. I was actually planning to purchase the specialization - just as a thank you gesture. Unfortunately, there is very little to be grateful for here.

In this course, you get a series of super short lectures giving you an overview of the math and physics behind the aerial robotics. I can't be the judge (since I hold a degree in this area), but I don't think you'll be able to get much if you aren't familiar with the topics yet. And you most certainly won't learn anything new if you had some previous training.

And then suddenly there are assignments. That have almost nothing to do with the lectures, but require quite a bit of Matlab programming (did I mention you won't be taught any Matlab in the process?) To make things worse, the assignments are rather poorly explained.

By the way, you won't get any replies from the course team - they even ignore questions about errors in lectures.

The course looks very sloppy. As if someone forced the team to put together something for Coursera. A huge disappointment :(

por Jeffrey

•Jan 16, 2016

The videos doesn't tell anything related to the test. How can I know the answers? Am I going to search them on the internet? Then what the meaning to learn this course?

por Md A S

•Feb 19, 2016

This is a totally theoretical course. This actually does not help to build a quadcopter from the scratch. It should have been mentioned in the introduction. I am opting out. The lectures did not seem appropriate for the people without proper background. Not satisfied.

por David C

•May 02, 2016

Very poor. Assignments consist of transcription of formulae given in lectures, and then manually searching the parameter space for solutions.

por Francesco B

•Jan 08, 2017

since there is a lot of math involved, there should be some carefully written readings. (not only slides or phd thesis)

por José I G C

•Feb 01, 2016

Really bad course. They don't explain the concepts clearly, the quizzes ambiguously ask things not seen in the videos and the instructor introduces equations without further explanation. The course *Autonomous Navigation for Flying Robots (edX) *is much better.

por David S

•May 25, 2016

Advertised as an easy course, week 1 is simple but quickly becomes far too difficult by week 2... Too late for your refund.

por Ran J

•Feb 28, 2016

The quiz and assignment are poorly designed. Important issues are overlooked or just briefly discussed in lecture. The assignment focuses on matlab coding, which only emphasizes the coding technique. Do not recommend to audience with

por Osama G M K

•Aug 09, 2016

This course is not even considered as an introduction it's a complementary for those whose career is robotics , in other words the required prerequisites are more than matlab programming skills and some calculus and algebra, its assignments have many problems that cannot be solved using just the course studying.

the course doesn't give many information.

por QUIJOUX

•Jul 04, 2017

Les thèmes abordés par le cours sot très intéressants, malheureusement le cours manque de structure. La difficulté n'est pas progressive et les exercices en ligne demandent une patience infini (solutions totalement empiriques, essai-erreur jusqu'à obtenir une valeur satisfaisante pour résoudre la solution). Il n'y a peu ou pas de soutien en ligne de la part de l'équipe ce qui fait que beaucoup d'élèves se retrouvent bloqués au dernier devoir. Une sensation de ne pas avoir appris grand chose à la fin de ce premier cours de robotique.

The topics of this course are really interesting but it has no internal consistency. The difficulty is not progressive and the online assignements require an infinite patience (the solutions are totally empirical and require multiple try & retry with the very slow simulation). There is more or less no online support from the team (thanks to the community) what leads to a big impediment for lots of students at the last assignment (cf. the discussion). Still have the feeling I did not learn much at the end of this first course on robotics.

por kavinstanes

•Mar 01, 2016

most difficult course

Coursera brinda acceso universal a la mejor educación del mundo, al asociarse con las mejores universidades y organizaciones, para ofrecer cursos en línea.