Title: “Continuous Locomotion with Stereo Fusion and Online Footstep Optimization.”
For humanoid robots to fulfill their potential for mobility, they must demonstrate reliable and efficient locomotion over rugged and irregular terrain, for example, with uneven and discontinuous surfaces. Here we present the perception and planning algorithms which have allowed a humanoid robot to walk over courses with these characteristics without stopping. We illustrate that passive stereo imagery can be used to measure the walking terrain with sufficient accuracy that a humanoid can accurately measure a terrain map using dense stereo fusion and then directly use it for footstep planning on the measured terrain. A particular integration challenge was ensuring that these two computationally intensive systems operate with minimal latency (below 1 second) to allow re-planning while walking.
Maurice Fallon is a Lecturer (Assistant Professor) in the School of Informatics at the University of Edinburgh. His research is focused on probabilistic methods for localization and mapping as well as motion planning for humanoid robotics. In particular, he is the perception lead on MIT’s Darpa Robotics Challenge team where he was first a Post Doc (2008-09) and then a Research Scientist until (2010-14). From 2004-2008 he completed his PhD studies at the University of Cambridge where he studied multi-target localization with acoustic sensors.