SC 635 – ADVANCED TOPICS IN MOBILE ROBOTICS

Course offered in:

2019-2020

Instructors:

Prof. Leena Vachhani, Prof. Arpita Sinha

Course Content:

The course was divided into 2 parts – single agent systems (upto midsem, taught by Prof. Leena Vachhani) and multi-agent systems (after midsem, taught by Prof. Arpita Sinha). There was a lab component for the course which involved using ROS and Gazebo to simulate different tasks to be performed by a mobile robot.

Topics covered before midsem:

  1. Kinematic modelling of mobile robots – task space, actuation space, non-holonomic constraints, unicycle model, differential drive wheeled mobile robots, car like wheeled mobile robots, etc.
  2. Localisation algorithms – Wavefront planner, Dijkstra’s algorithm, A* algorithm, Bug algorithms, etc.
  3. Probabilistic motion planning – belief distribution, Markov assumption, sensor measurements, etc.
  4. Kalman filter, extended kalman filter.

Very few lectures could be conducted after midsem due to suspension of semester but the aim was to extend the concepts for single agent systems to multi-agent systems.

Prerequisites: Must –

  1. Python programming
  2. Basic concepts in probability
  3. Basic concepts in linear algebra

Good to know –

  1. ROS + Gazebo

Previously “SC634 – Introduction to Mobile Robotics” used to be a prerequisite but it was removed this semester and it’s content was merged with SC635.

Feedback on Lectures:

This feedback is only for the pre-midsem part. There were theory lectures as well as lab lectures. 45 mins were devoted for lab lectures every week which were conducted by the TAs. The theory lectures were interesting but often difficult to follow. The instructor’s handwriting was barely legible which made it difficult to take notes. A lot of content was covered in a very short span of time. I often found that the instructor’s reasoning lacked mathematical rigor (maybe the intention was to let us do the hard math). Overall, I felt there was a lot of scope for improvement in teaching. The lab lectures were very moderately paced and easier to follow. I could understand a lot of theory just because of the laboratory component. The TAs were quite helpful.

Feedback on Tutorials, Assignments and Exams:

The Part 1 grading policy was as follows : Assignment 5%

Midsem 20%

Lab 5×4 = 20%

Viva 5%

There were 4 lab assignments in ROS+Gazebo based on theory covered in class. All of them were quite challenging but also fun to do. Extra sessions were conducted to clarify doubts related ROS. I found the midsem and assignment a bit tough. Viva was based on the lab work and reasonably simple questions were asked.

Difficulty:

(on a scale of 1 being very easy to 5 being very hard): 5

Grading Statistics:

AA – 3

AB – 7

BB – 9

BC – 3

CC – 3

FF – 1

S – 5

W – 1

Total – 32

Study Material and References:

  1. Probabilistic Robotics by S Thrun
  2. Introduction to Mobile Robot Control by Spyros G. Tzafestas

Takeaways from the course:

This is a very challenging course but if you are interested in the software side of mobile robotics you’ll definitely enjoy it.