PMID- 38202941
OWN - NLM
STAT- PubMed-not-MEDLINE
LR  - 20240113
IS  - 1424-8220 (Electronic)
IS  - 1424-8220 (Linking)
VI  - 24
IP  - 1
DP  - 2023 Dec 22
TI  - Knowledge Distillation for Traversable Region Detection of LiDAR Scan in Off-Road 
      Environments.
LID - 10.3390/s24010079 [doi]
LID - 79
AB  - In this study, we propose a knowledge distillation (KD) method for segmenting 
      off-road environment range images. Unlike urban environments, off-road terrains 
      are irregular and pose a higher risk to hardware. Therefore, off-road 
      self-driving systems are required to be computationally efficient. We used LiDAR 
      point cloud range images to address this challenge. The three-dimensional (3D) 
      point cloud data, which are rich in detail, require substantial computational 
      resources. To mitigate this problem, we employ a projection method to convert the 
      image into a two-dimensional (2D) image format using depth information. Our soft 
      label-based knowledge distillation (SLKD) effectively transfers knowledge from a 
      large teacher network to a lightweight student network. We evaluated SLKD using 
      the RELLIS-3D off-road environment dataset, measuring the performance with 
      respect to the mean intersection of union (mIoU) and GPU floating point 
      operations per second (GFLOPS). The experimental results demonstrate that SLKD 
      achieves a favorable trade-off between mIoU and GFLOPS when comparing teacher and 
      student networks. This approach shows promise for enabling efficient off-road 
      autonomous systems with reduced computational costs.
FAU - Kim, Nahyeong
AU  - Kim N
AUID- ORCID: 0009-0000-5878-9807
AD  - School of Computing, Gachon University, Seongnam-si 1332, Republic of Korea.
FAU - An, Jhonghyun
AU  - An J
AUID- ORCID: 0000-0003-3692-9754
AD  - School of Computing, Gachon University, Seongnam-si 1332, Republic of Korea.
LA  - eng
GR  - G21002236071/Technology development Program of MSS/
GR  - RS-2022-00165870/National Research Foundation of Korea(NRF) of MSIT/
GR  - 202300700001/Gachon University research fund of 2023/
PT  - Journal Article
DEP - 20231222
PL  - Switzerland
TA  - Sensors (Basel)
JT  - Sensors (Basel, Switzerland)
JID - 101204366
SB  - IM
PMC - PMC10781397
OTO - NOTNLM
OT  - LiDAR point cloud
OT  - knowledge distillation
OT  - off-road
OT  - point cloud projection
OT  - range image
OT  - self-driving
COIS- The authors declare no conflicts of interest.
EDAT- 2024/01/11 07:42
MHDA- 2024/01/11 07:43
PMCR- 2023/12/22
CRDT- 2024/01/11 01:15
PHST- 2023/11/10 00:00 [received]
PHST- 2023/12/18 00:00 [revised]
PHST- 2023/12/19 00:00 [accepted]
PHST- 2024/01/11 07:43 [medline]
PHST- 2024/01/11 07:42 [pubmed]
PHST- 2024/01/11 01:15 [entrez]
PHST- 2023/12/22 00:00 [pmc-release]
AID - s24010079 [pii]
AID - sensors-24-00079 [pii]
AID - 10.3390/s24010079 [doi]
PST - epublish
SO  - Sensors (Basel). 2023 Dec 22;24(1):79. doi: 10.3390/s24010079.