Robocup@Home 2018


Posted on July 5, 2018 at 12:49

Tag: Competition

Paper

As part of our lab's agreement with Toyota to lease their Human Support Robot (HSR), we have to showcase our research and participate in the Domestic Standard Platform League in Robocup@Home. All teams in the DSPL have the same HSR. Teams must qualify across 3 stages with each stage having its own challenges and qualifications. My role was to complete the Speech & Person Recognition task in the first stage and design the speech module that can be used by other robot components. (See my other post on Dialogflow & Google TTS).
Robots Rollout
Humans


Speech & Person Recognition Task

Despite rigorous testing prior to the competition, the robot was unaoble to answer most of the questions even though it had the capacity to. This was due to 1. the operators' heavy accents, 2. environment noise, and 3. no sound localization. I say sound localization because our robot is equipped with a microphone array but we did not have time to create sound localization software.
Humans

Main operator asking HSR initial questions

Human

Main operator asking HSR initial questions


Our team - Team Northeastern - was able to get 4th place out of 9 despite being the smallest team and having competed only once. Special thanks to my teammates Tarik Kelestemur and Naoki Yokoyama!
On-the-fly

Tarik demonstrating his on-the-fly grasping technique

Naoki & Tarik

Naoki and Tarik (LR) posing for a group photo

Qualification Video