Navigation Agents for the Visually Impaired: A Sidewalk Simulator and Experiments

Abstract

Millions of blind and visually-impaired (BVI) people navigate urban environments everyday, using smartphones for high-level path-planning and white canes or guide dogs for local information. However, many BVI people still struggle to travel to new places. In our endeavour to create a navigation assistant for the BVI, we found that existing Reinforcement Learning (RL) environments were unsuitable for the task. This work introduces SEVN, a sidewalk simulation environment and a neural network-based approach to creating a navigation agent. SEVN contains panoramic images with labels for house numbers, doors, and street name signs, and formulations for several navigation tasks. We study the performance of an RL algorithm (PPO) in this setting. Our policy model fuses multi-modal observations in the form of variable resolution images, visible text, and simulated GPS data to navigate to a goal door. We hope that this dataset, simulator, and experimental results will provide a foundation for further research into the creation of agents that can assist members of the BVI community with outdoor navigation.

Sim-to-Real Transfer for Vision-and-Language Navigation

“…For example, building on R2R/Matterport3D [13,21], annotations for vision-and-dialog navigation [35], asking for help [36], remote embodied referring expressions [37], and multilingual VLN [38,39] have been released. In the outdoor setting, several panoramic image datasets have been proposed including StreetLearn [40,41] and SEVN [42], giving rise to language navigation datasets such as TouchDown [43], Talk2Nav [44] and RUN [45]. With the increasing interest in training embodied agents in panoramic image environments, there is an urgent need to investigate the transfer of these agents to real physical platforms.…”