A Compressed Sensing Approach for Multiple Obstacle Localisation Using Sonar Sensors in Air.

Area of Science:

• Robotics
• Signal Processing
• Applied Physics

Background:

• Traditional autonomous navigation systems rely on sonar time-of-flight, which faces challenges with ultrasonic reflection interference.
• Multiple reflections in complex environments necessitate numerous acquisitions for accurate mapping.

Purpose of the Study:

• To present a novel approach for obstacle detection and localization in autonomous navigation.
• To address the limitations of traditional sonar methods using inverse problems and compressed sensing.

Main Methods:

• Utilized inverse problems and compressed sensing concepts for obstacle detection and localization.
• Conducted experiments with a custom hardware platform featuring four ultrasonic transducers.
• Compared compressed sensing (Orthogonal Matching Pursuit) with traditional image reconstruction methods.

Main Results:

• Successfully reconstructed 2D images representing the environment's cross-section.
• Demonstrated promising results for robotic mapping tasks compared to traditional methods.
• Showcased the effectiveness of compressed sensing in handling complex ultrasonic signal interference.

Conclusions:

• Compressed sensing offers a robust alternative to traditional sonar techniques for autonomous navigation.
• The proposed method enhances obstacle detection and localization accuracy in complex environments.
• This research paves the way for more efficient and reliable robotic mapping systems.