A hardware compressed sensing method for ultrasonic imaging

摘要

Ultrasonic sensing techniques have been widely used for autonomous robotic applications with low-profile and cost-effective ultrasonic distance sensors. These sensors are energy-effective and data-efficient as the power consumption is in the milliwatts-level and the output is typically ultrasonic time-of-flight (ToF) data. However, the wide angular directivity of small ultrasonic sensors causes large angular uncertainty for robotic sensing applications. This paper proposes a hardware compressed sensing method which can increase the sensing resolution for ultrasonic robotics and enables robotic exploration with ultrasonic imaging algorithms. A hardware modification is proposed for standard ultrasonic distance sensors to enable variable amplification for the received ultrasonic signals. Amplification-dependent ToF data can be acquired and treated as a series of Dirac delta functions in the time domain, which can then be represented as a hardware compressed signal and be used for ultrasonic imaging with an appropriate algorithm. Experiments showed that the proposed method can separate cluttered obstacles which cannot be identified by the robotic exploration approach with only ToF information. With the implementation of the synthetic aperture focusing (SAFT) method, the result shows that the proposed hardware compressed sensing method can distinguish multiple close obstacles using hardware compressed signals reconstructed using limited ultrasonic ToF data. • Innovative imaging: Uses limited ToF data for ultrasonic imaging with a new method. • Efficient data use: Converts few ToF measurements into a pseudo-RF signal with phase. • Cost-effective: Produces quality images without Nyquist sampling, reducing hardware needs.