Supervision Style and Guidelines

I supervise quite a lot of students in their (robotics-oriented) research projects (until now, at the master’s level), so I thought I would write this explanation of my supervision style (so that prospective students know what to expect from me), but also list some guidelines to make the collaboration easier. It should, however, be noted that I like to evolve my supervision style (in a way, every supervised project is a learning opportunity for me to grow as a supervisor); the notes of this page may change over time accordingly.

Supervision Style

My supervision style is characterised by (i) adapting to the profile of each student, (ii) giving research freedom to my students, and (iii) regularly providing experimental support. I elaborate on these aspects below.

• Personalised supervision: I typically try to adjust my supervision to the type of project I supervise and to the profile of the student. Most projects I supervise are initiated by me, so I have a clear idea about what I want to be achieved and thus want to accompany the work closely; in other words, I am more of a proactive supervisor in these projects. Occasionally, I also co-supervise projects that are not initiated by me, but in which a student has expressed an interest to work with me (e.g. due to my experience with similar prior work). In such projects, I tend to be a more reactive rather than a proactive supervisor, namely I don’t want to impose too much of myself in terms of the direction of the work (I expect this to be done by the main supervisor(s)); instead, I try to provide general guidance and simply be there for the student when they need feedback or support.
• Work freedom: As the projects I supervise are research-oriented (even if they are sometimes development-heavy), I like to provide as much freedom to my students as possible so that they can explore aspects of the work that are particularly interesting to them; I see the whole supervision process as a collaboration rather than a hierarchical supervisor-student relation in which a student only does what I tell them to do. As long as the work remains within the overall framework that we have agreed upon at the beginning of the project and aims to achieve the agreed upon general objectives, I am always open for new proposals made by my students and new directions in which the work can go. This means that I tend not to dictate what exactly students should do, and actually want them to be actively involved in defining the direction of the work. How much freedom I give does, however, depend on each student and is related to the above personalisation aspect; some students tend to work better when they have more direct guidance, while others thrive when they have more freedom; there is no golden rule that works equally well for everyone.
• Direct experimental support: As much as possible, I want to be involved in experimental work that is conducted in the context of the projects I supervise (particularly the ones that have been initiated by me). From my own work, I know that setting up experiments can be very time-consuming, so I want to support my students in this process as much as possible. In addition, again from my own work, I know that, with lack of experience, it becomes more likely that important steps in the process may be missed if students are left on their own; because of this, I try to share my experimental experience with students, thereby ensuring that experiments are designed correctly and that they cover all important aspects that we want to address with the study.

Expectations

I have a few general expectations from my students, in particular (i) willingness to learn new things, (ii) an ability to handle (potentially large bodies of) research literature, (iii) an ability to work with formal representations, and (iv) meeting deadlines. These are discussed in more detail below.

• Willingness to learn (and fail while learning): The obvious expectation I have from my students is that they are willing to learn as much as possible about the underlying topic of a project. For this, I expect some level of proactivity on the side of the student: projects progress best when students don’t always wait for me to tell them what to do, but instead take initiative to explore the topic on their own as well. Embracing failures is another essential aspect of the learning process; not everything that a student explores will work well, but that does not mean that they should be afraid to try new things out. After all, failing and then finding out why things have failed is the best way to learn something.
• Ability to read and evaluate scientific literature effectively: The work we do in research project is at the cutting edge, so it obviously involves reading a lot of (recent) scientific literature. This, in turn, means that I expect students to be able to read such literature:
  1. From the perspective of the project we are working on: Papers are never one-dimensional texts, namely there are always different ways in which they can be read and interpreted (concretely, they can be read from the point of view of the underlying representation, from the perspective of the use case, or by primarily considering the evaluation results for benchmarking). Keeping this in mind, I expect students to have an ability to focus on the aspects that are relevant for our work (and ignore the aspects that are not so relevant for us); this is the only way in which we can build on this work and improve upon it.
  2. Efficiently: Robotics is a fast-moving field, so there are always too many papers to read; because of this, I expect students to be able to read papers efficiently. What does this mean? In my experience, the best way to read efficiently is to perform multiple passes through a paper: once through the abstract to evaluate the general relevance for the work, then, for a relevant paper, another time by skimming through the whole paper, and then, for the papers that are most related to what we want to achieve, a more thorough read to fully understand it (but, again, by primarily focusing on the aspects that are relevant for the project — particularly relevant for long journal papers).
  3. With a critical eye: The scientific literature has (in my opinion unfortunately) a large positivity bias: the majority of research papers only focus on the positive aspects of the work, and limitations are not discussed or are only discussed briefly. Precisely because of this, it is essential to read every single paper with a critical eye (as a side note, I have written some of my thoughts on papers from a reviewer’s perspective here). This aspect is, however, also directly related to the part about reading from the perspective of our work — only limitations that are actually related to what we are trying to achieve are relevant to us, so these are the ones that we need to consider. Just looking for any limitation is not helpful at all, as the work may (also) have limitations that have nothing to do with our intended work.
• Ability to comprehend formal representations, or willingness to close knowledge gaps related to a lack of formal training: Based on my experience, one common reason why students sometimes struggle with making progress in research projects is that they aren’t comfortable with mathematical notation and formal representations. The majority of papers are, of course, not beginner-friendly, as they are written for experts and thus assume quite a bit of background on the side of the reader, but not understanding certain aspects of a paper should not discourage from trying to understand the content (for papers that are relevant to our work above all); instead, it should serve as motivation to backtrack a bit and attempt to cover the knowledge gap by consulting other resources that may explain some notation or a mathematical model better (this can include consulting other papers, books, blogs, forums discussions — anything may help). In addition, I really enjoy discussing formal representations, so I am particularly happy when a student approaches me to clarify doubts on that; in many cases, I also end up learning something new.
• Meeting deadlines: One of the most important aspects of a successful research project is solid time management, and meeting deadlines that I may set (e.g. for report drafts) is one obvious sign of good time management. Certainly, research progress is not always easy to predict, and life sometimes comes in the way, but if life comes in the way all the time and deadlines are missed without a prior discussion, then there is likely a time management or a procrastination problem. This is an issue that should be discussed openly (and early enough) so that the work can proceed successfully.

Communication Guidelines

When it comes to communication, I want to have regular meetings with my students, with a preference for face-to-face discussions. Taking notes during meetings is essential to prevent discussion points from being forgotten, and issues should be discussed early enough so that a project can be completed successfully.

• Regular meetings: Particularly in projects that have been initiated by me, I want to have regular progress meetings to ensure that progress is indeed happening and so that I can react to any concrete issues that a student is facing early enough. The frequency of these meetings tends to change during a project. In the early stages, progress is generally slower, as students usually need to spend more time reading relevant literature, so meetings are, by default, fortnightly (unless agreed upon differently with the student). As the project evolves and takes better shape, there is usually a need to discuss things more frequently, for instance to discuss concrete results and brainstorm improvements; at this point, meetings become weekly by default. I don’t have a hard rule on when the fortnightly meetings turn into weekly meetings, but this is usually after a month or two after the start of the project. For projects in which I am not the main supervisor, I prefer it if the frequency of meetings is dictated by the student; I want to be kept in the loop, but the student can decide whether we meet fortnightly, once a month, or at a different frequency.
• Preference for face-to-face meetings: When I started supervising several years ago, I used to prefer email-based communication with my students, particularly because I wanted more written content that I can give detailed feedback on (e.g. report drafts). Over time, I have learned to have a strong preference for face-to-face interaction, as we can resolve misunderstandings and discuss ongoing work much more directly. I still like looking at written drafts every once in a while, but my preference now is that we discuss those during a meeting (it is typically anyway the case that, given all of my other responsibilities — other than supervision I mean — I don’t have as much time for providing detailed feedback on written content as I used to).
• Taking meeting notes: Our memories are unreliable, but we tend to overestimate them quite frequently. The point is, even if everything seems clear during or right after a meeting, important details may be forgotten the more time passes on, particularly if the results of the discussion cannot be turned into action immediately. For this reason, I strongly encourage students to take notes from our meetings, while a meeting is taking place. How these notes are taken is not important — every student may have their own preferences; it is only important to keep them organised so that they are easy to navigate and refer to.
• Discussing issues openly and early enough: Research projects can be challenging for different reasons, such as the complexity of the subject itself, time constraints due to taking courses, or due to personal issues that one faces during the work. Any of these challenges can prevent progress in a project from happening and, while this is sometimes normal over a short term, the project as a whole will suffer if it is not put under control for a prolonged time. Thus, I want my students to feel free about openly discussing issues they might be facing in their work; in most cases, an open discussion will allow us to find a solution or take corrective action, for instance by adjusting the direction of the work or by changing the communication frequency. The worst thing that can happen in a project is that a student feels embarrassed to talk about a problem that could be resolved by a simple conversation.

So, that is a summary of my style as a supervisor, my expectations from students, as well as my communication preferences. I hope this has provided some clarity on what students can expect from me as a supervisor.