Vision-Based Uncertainty-Aware Motion Planning based on Probabilistic Semantic Segmentation

TL;DR

This paper introduces a vision-based motion planning method that uses probabilistic semantic segmentation with deep ensembles and data augmentation to improve safety and accuracy in uncertain environments.

Contribution

It presents a novel approach combining deep ensemble-based semantic segmentation with scenario-based planning for safer robot navigation under uncertainty.

Findings

Massive data augmentation improves segmentation reliability.

Deep ensembles provide accurate probabilistic occupancy information.

Scenario-based planning enhances safety despite perception uncertainties.

Abstract

For safe operation, a robot must be able to avoid collisions in uncertain environments. Existing approaches for motion planning under uncertainties often assume parametric obstacle representations and Gaussian uncertainty, which can be inaccurate. While visual perception can deliver a more accurate representation of the environment, its use for safe motion planning is limited by the inherent miscalibration of neural networks and the challenge of obtaining adequate datasets. To address these limitations, we propose to employ ensembles of deep semantic segmentation networks trained with massively augmented datasets to ensure reliable probabilistic occupancy information. To avoid conservatism during motion planning, we directly employ the probabilistic perception in a scenario-based path planning approach. A velocity scheduling scheme is applied to the path to ensure a safe motion despite tracking inaccuracies. We demonstrate the effectiveness of the massive data augmentation in combination with deep ensembles and the proposed scenario-based planning approach in comparisons to state-of-the-art methods and validate our framework in an experiment with a human hand as an obstacle.