Reproduce “GuideNav: User-Informed Development of a Vision-Only Robotic Navigation Assistant For Blind Travelers” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Hochul Hwang, Soowan Yang, Jahir Sadik Monon, Nicholas A Giudice, Sunghoon Ivan Lee, Joydeep Biswas, Donghyun Kim
GuideNav, a vision-only, teach-and-repeat robotic navigation assistant, was developed with user input to support blind and low-vision travelers. The system successfully demonstrated kilometer-scale autonomous guiding with BLV participants in varied outdoor environments, relying solely on an RGB camera.
Key insights
- User-centered design, informed by qualitative data from guide dog handlers and trainers, is critical for defining practical hardware and navigation requirements, such as optimal sensor suite, desired runtime, and adherence to Orientation and Mobility (O&M) principles.
- A vision-only, teach-and-repeat approach can achieve robust, kilometer-scale autonomous navigation for BLV individuals in varied outdoor environments, bypassing the cost, weight, and robustness issues associated with LiDAR and GPS.
- Direct relative camera pose regression models like ReLoc3r offer superior stability and accuracy for pose estimation in challenging outdoor conditions compared to traditional feature-matching and PnP methods, enabling smoother robot control.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.