Skip to content

Presenting Ask-to-Act at IEEE RO-MAN 2026 in Japan

· 2 min read · 0 views

Human-Robot InteractionRoboticsAI ResearchRO-MAN 2026

Last month, I traveled to Kitakyushu, Japan, to present my first-author paper (Ask-to-Act: Learning When Robots Should Clarify Ambiguous Grounded Instructions) at IEEE RO-MAN 2026. It wasn’t my first time in Japan or my first time presenting at a conference, but the experience was so much fun that it felt like I was doing both for the first time.

I also got to present three papers on behalf of my colleagues in the Interactive Robotics and Language Lab at the University of Maryland, Baltimore County. That made for a busy conference, but as you can imagine, I was never bored.

I should mention that my research looks at what a robot should do when someone gives it an instruction that could mean more than one thing. If the robot guesses, it may take the wrong action. But if it stops to ask a question every time, it quickly becomes frustrating to work with.

My aim was to figure out how a robot could decide when it has enough information to act and when it should ask for clarification first.

This is the direction I want to continue pursuing: robots that can follow instructions while also recognizing when they need more information to complete an assigned task.

I’m currently spending my weekends getting the repository ready to make public. I’ll share the repository and a link to the paper on my profile shortly for anyone interested in reading more.

It was rewarding to share the work, answer questions, and meet researchers working on similar problems from around the world. I’m grateful to my co-authors, advisor, and everyone who helped make the paper and this experience possible.

And while I got a lot of good work done while I was in Japan (which I’ll be posting about in the coming days), getting to experience more of the Land of the Rising Sun along the way made the trip even more memorable.

A bowl of Japanese shaved ice topped with dark red syrup and a swirl of soft-serve ice cream.

A traditional Japanese castle with white walls and layered tiled roofs, rising above a stone base.

Ekele standing beside a bronze sculpture of a figure holding birds aloft, on a waterfront promenade with a river, an arched bridge, and city buildings behind.

Ekele in a cap and backpack on a wide pedestrian bridge, with the castle and trees in the background.

Ekele wearing a conference badge beside the IEEE RO-MAN 2026 sign for Kitakyushu, Fukuoka, Japan, outside the convention center entrance.

A large conference auditorium with rows of wooden seats as attendees settle in for a session.

Ekele smiling at a wooden podium with a laptop, presenting next to a slide titled “The robot pays a cost either way”.

Ekele presenting at the podium in front of a slide explaining why clarification is hard: acting on a guess risks the wrong physical action, while asking every time turns a fluent interaction into an annoying one.

Originally posted on LinkedIn.