Course Information

Table of Contents

1) Course description

This course is designed as a variant of 6.100B that focuses on computational principles underlying autonomous systems. Our world increasingly relies on such systems, including for warehouse automation, self-driving/piloting vehicles, space and underwater operations, etc. To gain insight into how these systems work, we will cover fundamental modeling techniques and algorithms for making decisions. These include using graph search to plan, using probability to reduce uncertainty, leveraging optimization frameworks, and learning desired behavior from experience.

Compared to 6.100B, our topics are centered on decision-making rather than simulation and data science. However, we share the same goal of building confidence in programming for those who are relatively new to it, coming out of 6.100A or 6.100L. Exploring this track can help you preview if further courses in autonomy, such as 16.410 (principles of autonomy), 16.405/6.4200 (robotics lab), or 6.4110 (representation, inference, reasoning), would be of interest. In addition, taking this course will not limit you to robotics, as the techniques we cover are widely applicable in science, engineering, and even the arts.

2) Staff and contact

For most questions about course content or policies, please use intro-autonomy-staff@mit.edu, which will reach all instructors and TAs. Remember to CC the list when replying, too.

If you have a question of a more private or sensitive nature, and/or you would like to schedule a meeting, you may reach out to an instructor using their Kerberos listed below (append "@mit.edu").

Instructors

• Sertac Karaman (sertac)

Teaching assistants

• Juyeop Han (juyeop)
• Josef Biberstein (jxb)
• Lukas Lao Beyer (llb)

3) Class times and help

Our modes of instruction are through lectures, recitations, and office hours. These are available only in person; there will be no recordings or Zoom option. Our reasons for this are twofold:

• It's much easier to have meaningful discussions with you in person.
• Being a small class and a pilot offering, we prefer to focus our bandwidth on course content rather than on additional infrastructure.

3.1) Lectures

Lectures are on Mondays and Wednesdays 3–4:30 pm in 4-163. The first lecture is on Wednesday February 4 at 3:45–4:30 pm. These are the primary vehicle for introducing new material. (See details and policies on quizzes below.)

Finger exercises. After most lectures, we will release a short set of "finger exercise" problems to check your understanding on key concepts introduced in lecture. These exercises are designed to take only a few minutes. We highly recommend that you complete the finger exercises shortly after lecture, although there will be no set deadline (we just expect that you have completed all finger exercises, which make up 5% of your grade, by the end of the semester). Links to each lecture's finger exercises will be provided on the calendar on our home page.

3.2) Recitations

We have two recitation sections on Fridays. You may attend whichever section you prefer.

• 10 am in 32-124
• 1 pm in 32-124

Recitations are mainly to review and practice lecture material and will be optional. We may also give hints about the current week's problem set.

3.3) Office hours

On off-weeks, where there is no lecture, we will hold office hours during lecture times in 4-163 to help with understanding course material or completing problem sets.

Instructors are also available by appointment if you cannot make these times. Please email the staff list, and one of us will respond.

4) Assignments and grading

There are three categories of assignments, weighted as follows:

• 60% problem sets
• 35% quizzes
• 5% finger exercises

These comprise your total numeric score. However, we do not have pre-defined cutoffs for letter grades. Instead, we follow our best judgment to assign grades according to MIT's grade definitions. We strive to consider in whole the work you've submitted and the course staff's interactions with you.

4.1) Problem sets

We will release five problem sets on a biweekly basis that are each due in just under two weeks. These are programming tasks that exercise the modeling and algorithmic concepts from lecture on grounded scenarios.

We will assume you have installed 6.100's Python environment or equivalent. Please see their install instructions. Note, however, you can ignore their instructions on installing Python packages. The instructions for each pset will let you know what additional packages are needed, if any.

Regardless of your scores on all other components, in order to pass the class, you must submit every problem set in earnest.

4.2) Quizzes

There will be three quizzes held during certain lecture times. These are noted on the homepage calendar, and they correspond roughly to each preceding group of lectures.

Compared to a full-length midterm or final exam, quizzes are meant to give earlier feedback, but they still assess the depth of your understanding. Therefore, study for them as you would any other quiz, and take advantage of recitations and office hours to prepare.

Our quizzes will be closed-book, on paper, and hand-graded. Please remember to bring a pencil or pen. You may be asked to write brief Python code or to explain your reasoning in words. No notes or electronics are allowed, and we will provide any scratch paper.

5) Additional policies

5.1) Collaboration

Problem sets are designed to develop your programming skills. Thus, it is important that you come up with your own solutions and write your own code. You may discuss with others the problem statement to better understand the task at hand. However, any conversation that goes beyond specific input/output examples and towards general solutions is prohibited, except when talking to course staff. When we're helping you, we will guide your thinking in ways that still allow you to have ownership of the result.

These days, generative AI tools are readily available. While they have their value in the real world, using them to solve our programming assignments is detrimental to achieving the learning objectives of this class. Treat them as you would any other person in terms of collaboration.

Finally, it goes without saying that no communication or additional resources are allowed during quizzes. We take academic integrity violations on quizzes very seriously.

5.2) Pset extensions

There are no pre-allocated late days for psets. Instead, if you find yourself unable to submit when a pset is due, you may email the staff list on that day prior to the pset deadline (usually 10 pm), and we will grant you a 24-hour extension.

For any further extensions beyond this standard policy, we need to hear from Student Support Services (S³) that you've spoken with them about your situation, and that they would support additional flexibility.

5.3) Quiz make-ups

Because quizzes are during regular class time, and attendance is expected, we do not offer make-ups. We also do not drop any quiz score.

If you miss a quiz due to extraordinary circumstances, we will need to hear from S³ before considering a make-up. Note that even with S³ support, it is the instructors' decision whether to offer a make-up.

If you require testing accommodations, we need to receive confirmation from DAS by the end of the first week of classes. We will then contact you before each quiz with special instructions.