A photo I took while at ICRA 2022 in Philadelphia. This is the "Grand Hall" area where we had the conference reception. There are a lot of (more professional) photos on the conference website.

At long last, after more than two years of virtual conferences, last May I attended an in-person conference, the 2022 International Conference on Robotics and Automation (ICRA), from May 23-27. The last in-person conferences I attended were ISRR 2019 in Hanoi, Vietnam and NeurIPS 2019 in Vancouver, Canada (blog posts are here and here). Apologies for the massive months-long delay in blogging. One challenge with ICRA’s timing is that it was few weeks before the CoRL 2022 deadline, and so I (and many other attendees, as I would soon learn) were busy trying to work on our paper submissions.

Background and Context

ICRA is a large conference, held annually since 1984. You can find the list of past and future venues here. The last full in-person ICRA was in 2019 in Montreal, Canada. This year, it was in Philadelphia in a former train station converted to a large conference convention center. Philadelphia (or “Philly” as it’s often referred to in informal parlance) is also important in the political development of the United States and near the convention center are relevant museums and historical landmarks such as the Liberty Bell.

As with many other 5-day academic conferences and consistent with prior ICRAs, two are for workshops and three are for the main conference. Fortunately, I was able to attend the entire thing, plus stay an extra day after ICRA to explore Philly; call it a one-day vacation if you like.

I went to ICRA for several reasons. First, I wanted to return to an in-person conference experience. Second, it’s close: I work at CMU now, which is in the same state of Pennsylvania (albeit on the opposite side) so the travel isn’t too bad. Third, for the first time in my research career, I was a co-organizer of a workshop. Fourth, I also had a paper accepted which we called Planar Robot Casting, led by outstanding undergraduate researcher Vincent Lim and PhD student Huang (Raven) Huang. Yes, in case you’re wondering, we deliberately chose the PRC acronym because it matches the People’s Republic of China.

Another positive aspect of ICRA for me in particular is that ICRA and other IEEE conferences now have a new policy that explicitly allocates funding for compliance with the Americans for Disabilities Act (ADA). In 2020, my PhD advisors and I began asking the IEEE Robotics and Automation Society (RAS) to encourage them to provide funding for academic accommodations for all RAS-sponsored conference. Here’s a mid-2021 tweet by my PhD advisor Ken Goldberg summarizing what happened. Though he mentions me in his tweet, the real credit here goes to him and Torsten Kröger, and I thank both of them for their support. In the past, I would arrange for such academic accommodations to the conference by asking my university, but having the conference pay is probably the more appropriate outcome. Furthermore, while the impetus of this was my specific accommodation need, I hope this will extend for other accommodations.

I emailed ICRA a few months before the conference to get them to arrange for sign language interpreters. Rule of thumb: the process always takes longer than expected, so start early! After a long list of back-and-forth emails, ICRA was able to arrange for the services and it turned out quite well. Certainly, having the conference located in the United States was another plus. We’ll see if the process works out similarly well for ICRA 2023 in London, UK or ICRA 2024 in Yokohama, Japan. Hopefully the same staff members will be involved, as that might simplify the process considerably.

My Experience

On the first day, I arrived early to the convention center. The first reason was to meet the sign language interpreters, who often arrive early for such events. For me, it’s critical to also get there early so I can introduce myself to them.

The other reason for my early arrival was that I was one of the three lead co-organizers for the 2nd Workshop on Deformable Object Manipulation, along with Martina Lippi and Michael Welle. Those two were among the lead organizers last year and kindly invited me to help co-organize this year’s edition. Michael and I handled the in-person logistics for the workshop while Martina (in Italy) did a tremendous job holding the fort for all the remote needs.

We had a bit of a tight start to the workshop, because for some reason, either our room or our workshop was not properly registered with the conference and thus we didn’t have a microphone, a video projector, and other ingredients. Michael was able to hastily arrange for the conference staff to get the materials in time, but it was close. We also had issues with the microphone and audio in the morning and had to ask the convention staff to check. Other than the initial hiccups, I though the workshop went well. We had high attendance, especially for a workshop that was on the far end of the conference center. I’m guessing we regularly had 100 people in the room throughout the day (the exact number varied since people came in and out all the time). I met a lot of the authors and attendees, and I hope they all enjoyed the workshop as much as I enjoyed co-hosting and co-organizing it.

It was a nice experience, and I would be happy to organize a workshop again in the near future. I would write another blog post on “lessons learned” but this is still just my first time organizing so I don’t want to make any non-generalizable claims, and Rowan McAllister has already written a nice guide.

Later, we had the conference welcome reception in a larger space, where we could stand and grab food and drinks from various stalls. As usual, some of the food was distinctively local. We had Philadelphia-themed cheesesteaks which I tried and thought were good. These aren’t things I would normally eat, but it’s good to experience this once in my life. In fact I didn’t even realize Philly was known for cheesesteaks before coming to ICRA.

Then we had the next three conference days. I spent the mornings attending the major talks, and then the rest of the days doing my best to talk with as many people and groups as possible. Fortunately, it was much easier than I expected to snag impromptu conversations. I made a few “semi-formal” plans by emailing people to ask for meeting times, but for the most part, my interactions were highly random. I’m not sure if that’s the case for most people who attended the conference?

I tried to set my time so that my first priority was to talk with people. Then, my next priority was to explore the posters and company exhibits. If there was nothing else to do, then I would attend the smaller talks from authors of research papers. There were a few other events at ICRA that I recognized from prior years such as robots racing through small tracks. I don’t get too involved in such competitions as I worry they could detract from my research time.

Regarding communication and masks, there was a mix of people wearing their own masks and those eschewing masks. Most who wore masks were wearing N95 or other relatively high-quality masks. My communication with others at the conference definitely benefited enormously by having sign language interpreters, without which I would be constantly asking people to take off their masks when talking to me (I don’t like doing this).

The fifth and last day of the conference consisted of another day of workshops. It’s also one that I wasn’t able to experience too much. I spent most of the day working on my CoRL 2022 submission. I did this at the conference hotel due to the poor WiFi at the conference, but the hotel was very close by and could be accessible without walking outside. I also would later learn that other conference attendees were doing something similar as me. While “robot learning” is a subset of “robotics”, the CoRL community is growing in size (the 2022 edition had a record 505 submissions) and is an increasingly larger fraction of the overall ICRA community.

After the conference, I spent another full day in Philadelphia briefly exploring the nearby historical and political museums and landmarks. For example, I took a brief visit to the National Constitution Center.


A photo I took while inside the National Constitution Center after ICRA.

But, as mentioned earlier, I did not have too much spare time at my disposal. Thus, after briefly touring some of the city, I ordered a salad from Sweetgreen for dinner (at least they have them in Philly, unlike in Pittsburgh!), and to drink, I bought some low-sugar Boba from a small store in the Chinatown next to the convention center. Then, it was back to my hotel room to work on a paper submission. Since my flight was scheduled to leave very early the next morning, I opted to skip sleeping and just worked straight until around 3:30AM. Then I got an Uber ride to the airport and flew back to Pittsburgh.

Overall Impressions

Overall, I had a nice experience at ICRA 2022. When I look back at all the academic conferences that I have attended in my career, I clearly had the best networking experience I ever had at a conference. This was the first conference where I aimed from the beginning to spend less time attending talks and more time talking with people in smaller conversations. I also met a lot of people who I only knew from online interactions beforehand. I kept a document with a list of all these new interactions, and by the end of ICRA, I had met over 50 new people for the first time.

As always, there are some positives and negatives about the conference. In terms of the things that could be interpreted as negatives, a few come to mind:

  • For our workshop, we had some technical problems, such as our room being the only workshop room not initially assigned a video projector system as mentioned earlier. We also had the inevitable audio and microphone issues, and multiple attendees told us that they couldn’t hear speakers well. Also as expected, the hybrid format with online versus in-person attendees posed issues in that online speakers could often not hear speakers in the physical room. To clarify, besides the fact that our workshop and/or room didn’t seem to be properly registered at first, all these issues have been common in other conferences. I just wish there was a really easy way to get all the audio systems set up nicely. It seems like AV control systems are a universal headache.

  • There are strict union rules in Philadelphia which led to some surprises. For example, the conference convention employees arranged a set of poster boards for workshops, but if we moved those poster boards (e.g., across the hallway to be closer to a workshop room) which we did for our workshop since it was on the far end of a hallway, then the union could fine ICRA.

  • There seemed to be universal agreement among attendees that the food quality was not ideal. In my view, ICRA 2018 and 2019 had slightly better conference food offerings. This might be a downstream effect of COVID. Fortunately, there were a lot of food options in Philly near the convention center.

  • The WiFi was highly unreliable. This was problematic for us who were working on CoRL submissions. In fact, one conference attendee told me he was resorting to hand-writing (on his iPad) parts of his CoRL submission while at the conference (and then he would type it up in Overleaf in the hotel which had better WiFi). I did something similar by skipping most of the 5th day of ICRA to stay in my hotel to work on CoRL. Admittedly, the timing of CoRL was unfortunate, and this is not ICRA’s fault.

  • COVID isn’t actually over, and there are likely new variants circulating. I don’t know how many of the conference attendees tested positive afterwards. Whatever the case, I think it remains imperative for us to watch for the health of the community and to try and arrange for conference venues that will maximize the amount of physical space available. I know this is not easy for a conference like ICRA. During some of the dinners and receptions, many people were packed in tight quarters. In-person conferences will continue so I hope the community will develop best practices for mitigating infections. Despite the many technical challenges with combining in-person and virtual experiences, I think conferences should continue offering virtual options and not require paper authors to attend.

And yet … when I look at what I wrote above, I almost feel like these are nitpicks. Putting together a massive conference with thousands of people is incredibly complex with hundreds of administrative tasks that have to be done, and getting something that works for everyone is impossible. I sincerely appreciate and respect all the hard work that has to be done to execute a conference of this magnitude. In the end, maybe we should just focus on how fortunate we are that we can get together for a conference? Look at all the excitement in the photos from the conference here. Academic life is attractive for a reason, and so many of us (myself included for sure) are so fortunate that we can be a part of this exciting field. I can’t wait to see how robotics evolves going forward, and I hope many others share this excitement.