A Dual-Field Sensing Scheme for a Guidance System for the Blind.

Area of Science:

• Robotics and Human-Computer Interaction
• Computer Vision and Sensor Fusion
• Assistive Technologies

Background:

• Electronic guidance systems aid visually impaired individuals in local environmental perception.
• Previous work proposed a context-aware system using laser scanners and cameras for near-field scene interpretation.
• High-resolution near-field scene interpretation was achieved using a graphical model.

Purpose of the Study:

• To extend previous work by incorporating a far-field graphical model.
• To develop a dual-field sensing scheme integrating near-field and far-field data.
• To optimize scene interpretation capability for guidance systems with simple sensor configurations.

Main Methods:

• Integration of near-field and far-field graphical models into a dual-field sensing scheme.
• Fusing range and image data using a near-field graphical model for ground and object status inference.
• Utilizing a far-field graphical model to interpret camera-based appearance and spatial prototypes from near-field data.

Main Results:

• The dual-field sensing scheme effectively interprets both near-field and far-field environmental data.
• Reliable inference of ground and object status in the near-field was achieved.
• Far-field image data interpretation was enhanced using prototypes derived from near-field data.

Conclusions:

• The proposed dual-field sensing scheme significantly improves blind people's perception in urban environments.
• This approach offers an effective solution for optimizing scene interpretation in guidance systems.
• Simple sensor configurations can achieve advanced scene interpretation capabilities.