C-BRIC Researchers to Present at ICRA

C-BRIC research will be well represented at the 2022 IEEE International Conference on Robotics and Automation (ICRA). The conference will be in Philadelphia, Pennsylvania, May 23-27, 2022.
Congratulations to researchers Aqeel Anwar, Adarsh Kosta, Vijay Kumar, Laura Jarin-Lipschitz, Chankyu Lee, Xu Liu, Priya Panda, Arijit Raychowdhury, Kaushik Roy, and Yuezhan Tao.
Chankyu Lee, Adarsh Kosta, and Kaushik Roy’s work on “Fusion-FlowNet: Energy-Efficient Optical Flow Estimation using Sensor Fusion and Deep Fused Spiking-Analog Network Architectures” will be presented at ICRA. This work presents a sensor fusion framework for energy-efficient optical flow estimation using frame- and event-based sensors. The network used is a fusion of Spiking Neural Networks (SNNs) and Analog Neural Networks (ANNs). Lee, a Purdue University PhD graduate (now working at Intel), performed Fusion-FlowNet research with C-BRIC. Kosta is a current PhD student at Purdue. Lee and Kosta both worked under the direction of Professor Kaushik Roy.
Adarsh K. Kosta, Aqeel Anwar, Priya Panda, Arijit Raychowdhury, and Kaushik Roy will present their work on “RAPID-RL: A Reconfigurable Architecture with Preemptive-Exits for Efficient Deep-Reinforcement Learning” at ICRA. RAPID-RL proposes a reconfigurable architecture with preemptive exits for efficient deep RL, which dynamically adjusts the compute effort during inference while maintaining competitive performance. Anwar is a PhD graduate of the Georgia Institute of Technology and currently works at NVIDIA. Anwar conducted the research for this paper during his time with C-BRIC; under the direction of Professor Arijit Raychowdhury. Kosta and Roy are C-BRIC researchers at Purdue University and Professor Priya Panda is a C-BRIC researcher at Yale University.
Laura Jarin-Lipschitz, Xu Liu, Yuezhan Tao, and Vijay Kumar will present work on “Experiments in Adaptive Replanning for Fast Autonomous Flight in Forests” at ICRA. This paper presents the work of the group’s planning framework with two parts: one that maximizes planner completeness for a given graph size and a second that dynamically maximizes graph size subject to computational constraints. This method allows operation in environments with varying density. Jarin-Lipschitz, Lui, and Tao are all current PhD students under the direction of Professor Vijay Kumar.