Road Environment Recognition for Automotive FMCW RADAR Systems Through Convolutional Neural Network

Sim, Heonkyo; Do, The-Duong; Lee, Seongwook; Kim, Yong‐Hwa; Kim, Seong-Cheol

doi:10.1109/access.2020.3013263

Cited by 13 publications

(6 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the context of research for using Doppler enabled sensors for positioning, [28] presents an approach where a rotating FMCW radar is used to estimate ego-motion and build a map based on the static returns. In [27] and [11], the FMCW radar spectral information is used for place and pose estimation. [15] proposed a Doppler velocity-based cluster and velocity estimation algorithm using an FMCW LiDAR.…”

Section: Related Workmentioning

confidence: 99%

DICP: Doppler Iterative Closest Point Algorithm

Bruno¹,

Vhavle²,

Chen³

2022

Preprint

View full text Add to dashboard Cite

In this paper, we present a novel algorithm for point cloud registration for range sensors capable of measuring perreturn instantaneous radial velocity: Doppler ICP. Existing variants of ICP that solely rely on geometry or other features generally fail to estimate the motion of the sensor correctly in scenarios that have non-distinctive features and/or repetitive geometric structures such as hallways, tunnels, highways, and bridges. We propose a new Doppler velocity objective function that exploits the compatibility of each point's Doppler measurement and the sensor's current motion estimate. We jointly optimize the Doppler velocity objective function and the geometric objective function which sufficiently constrains the point cloud alignment problem even in feature-denied environments. Furthermore, the correspondence matches used for the alignment are improved by pruning away the points from dynamic targets which generally degrade the ICP solution. We evaluate our method on data collected from real sensors and from simulation. Our results show a significant performance improvement in terms of the registration accuracy with the added benefit of faster convergence guided by the Doppler velocity gradients.

show abstract

Section: Related Workmentioning

confidence: 99%

DICP: Doppler Iterative Closest Point Algorithm

Bruno¹,

Vhavle²,

Chen³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…When applying the Fourier transform [ 25 ] to the signal of Equation ( 1 ), we can obtain a baseband signal in the frequency domain, which can be expressed as

where

is the frequency index. By accumulating the frequency-domain baseband signal over time, we can obtain the spectrogram, which shows the change in distance of an object over time [ 26 ]. In other words, in the FMCW radar system, the detection result in the frequency domain can be interpreted as that in the distance domain [ 24 ].…”

Section: Radar Sensor Data Acquisition and Signal Preprocessingmentioning

confidence: 99%

“…, K − 1) is the frequency index. By accumulating the frequencydomain baseband signal over time, we can obtain the spectrogram, which shows the change in distance of an object over time [26]. In other words, in the FMCW radar system, the detection result in the frequency domain can be interpreted as that in the distance domain [24].…”

Section: Radar Sensor Data Acquisition and Signal Preprocessing 21 Fmcw Radar Sensormentioning

confidence: 99%

Identification of Human Motion Using Radar Sensor in an Indoor Environment

Kang

Jang

Lee

2021

Sensors

Self Cite

View full text Add to dashboard Cite

In this paper, we propose a method of identifying human motions, such as standing, walking, running, and crawling, using a millimeter wave radar sensor. In our method, two signal processing is performed in parallel to identify the human motions. First, the moment at which a person’s motion changes is determined based on the statistical characteristics of the radar signal. Second, a deep learning-based classification algorithm is applied to determine what actions a person is taking. In each of the two signal processing, radar spectrograms containing the characteristics of the distance change over time are used as input. Finally, we evaluate the performance of the proposed method with radar sensor data acquired in an indoor environment. The proposed method can find the moment when the motion changes with an error rate of 3%, and also can classify the action that a person is taking with more than 95% accuracy.

show abstract

“…Frequency Modulated Continuous Wave radars are inexpensive and all-weather, and have served as the key sensor for modern ADAS. Ongoing advances are improving radar resolution and target discrimination [14], while convolutional networks has been used to add discriminative power to radar data, moving beyond target detection and tracking to include classifying road environments [24,41], and seeing beyond-line-of-sight targets [40]. Nevertheless, the low spatial resolution of radar means that the 3D environment, including object shape and classification, are only coarsely obtained.…”

Section: Related Workmentioning

confidence: 99%