FPGA Accelerators for Robust Visual SLAM on Humanoid Robots

The paper on "FPGA Accelerators for Robust Visual SLAM on Humanoid Robots" will be presented as a poster on the 30th ACM/SIGDA International Symposium on Field-Programmable Gate Arrays (FPGA). This work presents a pose graph optimization module as an extension on the KinectFusion pipeline (a well-known SLAM algorithm), to help recover the robot’s pose during unstable gait patterns where the KinectFusion tracking system fails. The work implements a number of precise and approximate optimizations to achieve real-time performance on the ZCU102 FPGA hardware.

This work is in cooperation with the Foundation for Research and Technology - Hellas (FORTH).