Georgia Tech Showcases Robotics Research at ICRA 2015

Georgia Tech’s Institute for Robotics and Intelligent Machines (IRIM) has a strong showing at ICRA, the 2015 IEEE International Conference on Robotics and Automation. Throughout the week, our researchers will present 28 papers and 4 posters, as well as provide leadership through chairing sessions, holding workshops and forums, and participating on technical panels.

SATURDAY

Today marked the final day of ICRA 2015 where Sugmoon Joo presented a paper on Using the Hubo Platform to Advance Humanoids Research in a workshop devoted to the Hubo platform.

Samantha Krening, a graduate student who works with Karen Feigh, Andrea Thomaz, and Charles Isbell, presented Human Advice in Object-Focused Reinforcement Learning during the poster session of the Becoming a Robot Guru: Integrating Science, Engineering and Creativity workshop.

In the Autonomy for Aquatic Robotics: the Role of Control and Learning in Single and Multi-Robot Systems​ workshop, Fumin Zhang presented a paper on Persistent Controlled Lagrangian Prediction Theory and Michael “Misha” Novitzky/Tucker R. Balch​ presented a poster on Towards Trajectory Adaptation for Recognition​.

Heni Ben Amor gave a talk on Robot Learning for Grasping and Manipulation as part of the workshop on Robotic Hands, Grasping, and Manipulation.

FRIDAY

Friday has been a busy day for IRIM researchers at ICRA 2015.​ Henrik Christensen chaired a session on Grasping, while Frank Dellaert chaired two sessions—one on SLAM and another on Aerial Robots Navigation.

Faculty members and students presented 13 papers today, representing research conducted in the labs of Ayanna Howard, Henrik Christensen, Daniel Goldman, Charles Isbell, Patricio Vela, Kok-Meng Lee, James Rehg, Panagiotis Tsiotras, Frank Dellaert, and Zsolt Kira.

  • Duality-Based Verification Techniques for 2D SLAM (Luca Carlone/Frank Dellaert​)
  • Initialization Techniques for 3D SLAM: A Survey on Rotation Estimation and Its Use in Pose Graph Optimization​ (Luca Carlone/Frank Dellaert​)
  • An Evaluation of Features for Classifier Transfer During Target Handoff across Aerial and Ground Robots (Zsolt Kira​​)
  • Information Based Reduced Landmark SLAM (Siddharth Choudhary​/Henrik Christensen & Frank Dellaert​​)
  • Multi-scale Perception and Path Planning on Probabilistic Obstacle Maps (Florian Hauer & Abhijit Kundu/James RehgPanagiotis Tsiotras​)
  • Optimally Observable and Minimal Cardinality Monocular SLAM (Guangcong Zhang/Patricio Vela​)
  • Dynamic Programming Guided Exploration for Sampling-Based Motion Planning Algorithms (Oktay Arslan/Panagiotis Tsiotras)
  • Design Analysis of a Passive Weight-Support Lower-Extremity-Exoskeleton with Compliant Knee-Joint (Kok-Meng Lee)
  • Learning Non-Holonomic Object Models for Mobile Manipulation (Jonathan Scholz & Martin Levihn/Charles Isbell & Henrik Christensen)
  • Distributed Real-Time Cooperative Localization and Mapping Using an Uncertainty-Aware Expectation Maximization Approach (Jing Dong/Frank Dellaert​​)
  • Robot-Inspired Biology: The Compound-Wave Control Template (Henry Astley/Daniel Goldman)
  • Monocular Image Space Tracking on a Computationally Limited MAV (Frank Dellaert​​)
  • Retrieving Experience: Interactive Instance-Based Learning Methods for Building Robot Companions (Hae Won Park/Ayanna Howard)

THURSDAY

The Amazon Picking Challenge results are in. Although the Georgia Tech team did not win an award (given only to the top three teams), they performed remarkably well. Out of 25 teams that competed in the Challenge, GT tied for 10th place overall, surpassing the median score. Even more impressive, 12 teams finished with a score of 0 and five were unable to coach their robot to physically move, disqualifying them from the competition. 

Patricio Vela​​ chaired the session on Bio-Inspired Robotics: Snakes and Climbing and Henrik Christensen participated in the Government Forum panel.

Both Henrik Christensen and Spiridon Reveliotis were recognized as members of the 2015 class of IEEE Robotics and Automation Society Fellows at the ICRA awards ceremony. Also, Learning Multiple Collaborative Tasks with a Mixture of Interaction Primitives, a paper co-authored by Heni Ben Amor, was recognized as a best paper finalist in the service robotics category. Stephen Balakirsky, Competitions Committee chair for the IEEE Robotics and Automation Society, presented the Robot Challenge Awards.

IRIM researchers presented more papers, representing research conducted in the labs of Karen Liu, Henrik Christensen, Patricio Vela, Magnus Egerstedt, and Daniel Goldman.

  • Reducing Hardware Experiments for Model Learning and Policy Optimization (Sehoon Ha​​/Karen Liu)
  • Incorporating Frictional Anisotropy in the Design of a Robotic Snake through the Exploitation of Scales (Miguel Serrano, Alexander Chang & Guangcong Zhang/Patricio Vela​)
  • Exploiting Symmetries and Extrusions for Grasping Household Objects (Ana Huamá​n Quispe, Benoît Milville, Can Erdogan, Heni Ben Amor & Mike Stilman☨/Henrik Christensen)
  • The GRITSBot in Its Natural Habitat—A Multi-Robot Testbed (Daniel Pickem & Myron Lee/Magnus Egerstedt​)
  • Limbless Locomotors That Turn in Place (Henry Astley/Daniel Goldman)

Additionally, students presented two posters:

  • Localization of Autonomous Underwater Vehicles Incorporating Flow Models and Acoustic Detection​ (Sungjin Cho/Fumin Zhang)
  • Learning Spatio-Temporal Features of Prompting During Robot Intervention for Children with Autism (Bi Ge & Hae Won Park/Ayanna Howar​d)

WEDNESDAY

Jun Ueda chaired the Vision: Detection, Recognition and Segmentation session and Ayanna Howard chaired the Radhika Nagpal’s keynote talk, “Taming the Swarm.” IRIM researchers presented multiple papers, representing research conducted in the labs of Aaron Bobick, Henrik Christensen, Frank Dellaert, Charlie Kemp, and Jun Ueda.

  • Antagonistic Muscle Based Robot Control for Physical Interactions (Tapomayukh Bhattacharjee​/Charlie Kemp)
  • Augmenting Physical State Prediction through Structured Activity Inference (Nam Vo/Aaron Bobick)
  • Differential Dynamic Programming for Optimal Estimation (Duy-Nguyen Ta/Frank Dellaert)
  • Learning Multiple Collaborative Tasks with a Mixture of Interaction Primitives (Heni Ben Amor/Henrik Christensen)
  • Real-Time Image De-Blurring and Image Processing for a Robotic Vision System (Michael D. Kim/Jun Ueda)

Ph.D. student Eric Huang, together with three undergraduate students, competed in the first ever Amazon Picking Challenge using Crichton the robot to pick objects off warehouse shelves. Congratulations are in order for the white and gold team! Since arriving in Seattle, none of them have slept, and they worked through the night to reassemble Crichton and construct a stable base for the robot. Despite many logistical challenges, preliminary results show the Georgia Tech team ranked in 6th place out of 28 teams. Go Jackets!

TUESDAY

Byron Boots presented Advances in Sensorimotor Learning and Ayanna Howard chaired the Ph.D. Forum. Hae Won Park, a post-doctoral researcher in Howard’s HumAnS Lab, served as one of the co-chairs of the Ph.D. Forum.