System, method, and computer program product for obstacle distance detection

Abstract

This disclosure provides an obstacle distance detection system. The processor receives two radar echoes and includes a separation module, a filtering module, a waveform remodeling module, and a distance estimation module. The separation module uses an independent component analysis algorithm to separate the radar echoes into two separate signals. The filtering module uses a recursive least squares filter to filter the two separate signals and outputs a correction signal. The waveform remodeling module analyzes the correction signal using a variational mode decomposition algorithm to obtain complex intrinsic mode functions (IMFs) and performs spectral transformation. Based on the spectral transformation result, one of the complex IMFs is removed to obtain the reconstructed denoised echo signal. The distance estimation module calculates the distance to the obstacle using the denoised echo signal, thereby improving the accuracy of distance estimation.

Claims

1. An obstacle distance detection system, comprising: a vehicle; two radars spaced apart from the vehicle, each radar emitting a transmitted wave and receiving a radar echo reflected by an obstacle; and a processor disposed on the vehicle and signal-connected to the two radars, the processor receiving the radar echoes and comprising: a separation module performing signal separation on the two radar echoes using an independent component analysis algorithm to separate two separate signals corresponding to the two radar echoes respectively; a filtering module filtering the two separate signals using a recursive least squares filter and outputting a correction signal; a waveform remodeling module analyzing the correction signal using a variational mode decomposition algorithm to obtain complex intrinsic mode functions and perform a spectrum transformation, removing one of the intrinsic mode functions based on a result of the spectrum transformation to obtain a reconstructed denoised echo signal; and a distance estimation module calculating a distance to the obstacle using the denoised echo signal.

2. The obstacle distance detection system as described in claim 1, wherein the waveform remodeling group performs the spectral conversion using a Hilbert-Huang conversion.

3. The obstacle distance detection system as described in claim 1, wherein the number of the intrinsic mode functions in the waveform remodeling group is set to five.

4. The obstacle distance detection system as described in claim 1, wherein each of the two radars is a frequency-modulated continuous wave radar.

5. An obstacle distance detection method, comprising: a radar detection step, wherein two radars mounted on a vehicle detect an obstacle, each of the two radars emitting a transmitted wave and each of the two radars receiving a radar echo formed by reflection from the obstacle; a separation step, wherein a processor mounted on the vehicle and signal-connected to the two radars receives the two radar echoes, and a separation module of the processor performs signal separation on the two radar echoes using an independent component analysis algorithm to separate two separate signals corresponding to the two radar echoes respectively; and a filtering processing step, wherein a filtering processing module of the processor filters the two separate signals using a recursive least squares filter and outputs a correction signal. A waveform reconstruction step, wherein a waveform remodeling group of the processor analyzes the correction signal using a variational mode decomposition algorithm, obtains complex intrinsic mode functions and performs a spectrum transformation, and removes one of the intrinsic mode functions based on the result of the spectrum transformation to obtain a reconstructed denoised echo signal; and a distance estimation step, wherein a distance estimation module of the processor calculates a distance to the obstacle using the denoised echo signal.

6. The obstacle distance detection method as described in claim 5, wherein the waveform remodeling group performs the spectral conversion using a Hilbert-Huang conversion.

7. The obstacle distance detection method as described in claim 5, wherein the number of the intrinsic mode functions in the waveform remodeling group is set to five.

8. A computer program product that enables a computer to analyze the distance to an obstacle, the computer receiving two radar echoes from two radars, and the computer program product causing the computer to perform the following: performing an independent component analysis algorithm to separate the two radar echoes into two separate signals corresponding to the two radar echoes; filtering the two separate signals with a recursive least squares filter and outputting a correction signal; performing a variational mode decomposition algorithm to analyze the correction signal, obtain complex intrinsic mode functions, perform a spectrum transformation, remove one of the intrinsic mode functions based on a result of the spectrum transformation to obtain a reconstructed denoised echo signal; and calculating the distance to the obstacle using the denoised echo signal.

9. The computer program product as described in claim 8, wherein a Hilbert-Huang converter is used to perform the spectrum conversion.

10. The computer program product as described in claim 8, wherein the number of the intrinsic modal functions is five.

Images