Weakly-Supervised Recommended Traversable Area Segmentation Using Automatically Labeled Images for Autonomous Driving in Pedestrian Environment with No Edges

Abstract

Detection of traversable areas is essential to navigation of autonomous personal mobility systems in unknown pedestrian environments. However, traffic rules may recommend or require driving in specified areas, such as sidewalks, in environments where roadways and sidewalks coexist. Therefore, it is necessary for such autonomous mobility systems to estimate the areas that are mechanically traversable and recommended by traffic rules and to navigate based on this estimation. In this paper, we propose a method for weakly-supervised recommended traversable area segmentation in environments with no edges using automatically labeled images based on paths selected by humans. This approach is based on the idea that a human-selected driving path more accurately reflects both mechanical traversability and human understanding of traffic rules and visual information. In addition, we propose a data augmentation method and a loss weighting method for detecting the appropriate recommended traversable area from a single human-selected path. Evaluation of the results showed that the proposed learning methods are effective for recommended traversable area detection and found that weakly-supervised semantic segmentation using human-selected path information is useful for recommended area detection in environments with no edges.

TrackletMapper: Ground Surface Segmentation and Mapping from Traffic Participant Trajectories

“…Zürn et al [13] propose a self-supervised labeling scheme based on an unsupervised audio clustering approach, where the cluster indices serve as weak labels and are projected into the robot camera images. Most recently, Onozuka et al [11] propose a traversable area segmentation approach for personal mobility systems such as intelligent wheelchairs.…”

“…Automatic annotation of images offers a promising alternative to manual annotations made by human annotators. Previously proposed automatic annotation approaches typically leverage the ego-motion of a data collection platform to obtain spatially sparse image-level labels of traversable ground surfaces [10,11] or are based on proprioceptive sensors such as sound and vibration [12,13,14]. In contrast to existing work, we additionally leverage the trajectories of other traffic participants such as vehicles and pedestrians, and project them into the camera images.…”

A Comparison Study of Neural Network-Based Semantic Segmentation for Off-Road Traversability

“…The method was validated using an ANYmal quadruped robot in unstructured environment including terrain types such as asphalt, dirt, sand, grass, and also considering different weather and light conditions. From a more human-driven knowledge viewpoint, work in Onozuka et al [3] proposed an automatic labelling system based on human-driven knowledge. A twostep approach was followed i.e.…”