seminars

Vojtěch Šalanský presents Pose consistency KKT-loss for weakly supervised learning of robot-terrain interaction model

On 2021-06-01 11:00 at https://feectu.zoom.us/j/96604651389
We address the problem of self-supervised learning for predicting the shape of
supporting terrain (i.e., the layer of terrain which provides rigid support for
the robot during its traversal) from sparse lidar measurements. The learning
method exploits two types of ground-truth labels: dense 2.5D maps and robot
poses, both estimated by a usual SLAM procedure from offline recorded
measurements. We claim and experimentally show that the robot poses are
required because the straightforward supervised learning from the 2.5D maps only
suffers from (i) exaggerated height of the supporting terrain caused by terrain
flexibility (vegetation, shallow water, snow, or sand) and (ii) missing or noisy
measurements caused by high spectral absorbance or non-Lambertian reflectance of
the measured surface. We address the learning from robot poses by introducing a
novel KKT-loss, which emerges as the distance from necessary Karush-Kuhn-Tucker
conditions for constrained local optima of a simplified first-principle model of
the robot-terrain interaction. We experimentally verify that the proposed weakly
supervised learning from ground-truth robot poses boosts the accuracy of
predicted support heightmaps and increases the accuracy of estimated robot
poses. All experiments are conducted on a dataset captured by a real platform.
Both the dataset and codes which replicate experiments in the paper are made
publicly available as a part of the submission.

Pose consistency KKT-loss for weakly supervised learning of robot-terrain
interaction model V Salansky, K Zimmermann, T Petricek, T Svoboda - IEEE
Robotics and Automation Letters, 2021

Responsible person: Petr Pošík