- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
[IEEE 2019 IEEE Intelligent Transportation Systems Conference - ITSC - Auckland, New Zealand (2019.10.27-2019.10.30)] 2019 IEEE Intelligent Transportation Systems Conference (ITSC) - Training a Fast Object Detector for LiDAR Range Images Using Labeled Data from Sensors with Higher Resolution
摘要： In this paper, we describe a strategy for training neural networks for object detection in range images obtained from one type of LiDAR sensor using labeled data from a different type of LiDAR sensor. Additionally, an efficient model for object detection in range images for use in self-driving cars is presented. Currently, the highest performing algorithms for object detection from LiDAR measurements are based on neural networks. Training these networks using supervised learning requires large annotated datasets. Therefore, most research using neural networks for object detection from LiDAR point clouds is conducted on a very small number of publicly available datasets. Consequently, only a small number of sensor types are used. We use an existing annotated dataset to train a neural network that can be used with a LiDAR sensor that has a lower resolution than the one used for recording the annotated dataset. This is done by simulating data from the lower resolution LiDAR sensor based on the higher resolution dataset. Furthermore, improvements to models that use LiDAR range images for object detection are presented. The results are validated using both simulated sensor data and data from an actual lower resolution sensor mounted to a research vehicle. It is shown that the model can detect objects from 360? range images in real time.
关键词： self-driving cars,object detection,LiDAR,range images,neural networks
更新于2025-09-19 17:13:59
-
[IEEE 2019 7th International Conference on Smart Computing & Communications (ICSCC) - Sarawak, Malaysia, Malaysia (2019.6.28-2019.6.30)] 2019 7th International Conference on Smart Computing & Communications (ICSCC) - On Self Driving Cars: An LED Time of Flight (ToF) based Detection and Ranging using various Unipolar Optical CDMA Codes
摘要： The dramatic surge in the development of autonomous vehicles has generated a need for, among others, improving detection accuracy. Various technologies and techniques have been proposed and adopted for autonomous vehicles, such as computer vision, in conjunction with machine learning and deep learning – neural network. Despite many attempts to improve the accuracy of detection and ranging, the computational load in training the neural network due to the requirement of massive amount of data had increased to a significant extent, along with the cost of operation. Light Detection and Ranging (LiDAR) systems are a new approach to improve the detection and ranging accuracy. However, LED Detection and Ranging (LEDDAR) based systems have not been explored so far. Herein, we present an analysis of the method of detection and ranging using LED and determine its performance for different number of users using Unipolar Optical Code Division Multiple Access (OCDMA) Codes. This was done using probability of error vs. Signal – to – Noise Ratio (SNR) and Distance (in m) using misdetection and false alarm analysis for a given transmitted power of LED obeying the current standards. The method of LED based beamforming by deploying multiple LED’s in an array is proposed to improve the detection and ranging accuracy. Finally, the results show that OCDMA based OOC codes show a low probability of error for a given SNR and Distance and outperformed other OCDMA techniques such as prime codes and hence indicate that OOC codes will be an optimal choice that can be coded in LED’s for the use in self driving cars.
关键词： Computer Vision,Neural Network,Misdetection,Signal – to – Noise Ratio,Beamforming,Probability of Error,False Alarm,Self – Driving Cars,Field – of - View
更新于2025-09-12 10:27:22
-
[IEEE 2018 IEEE International Conference on Intelligent Transportation Systems (ITSC) - Maui, HI, USA (2018.11.4-2018.11.7)] 2018 21st International Conference on Intelligent Transportation Systems (ITSC) - Automotive LiDAR performance verification in fog and rain
摘要： This article focuses on testing and investigating further development needs for LiDARs in self-driving cars in adverse weather. The article compares two different LiDARs (Ibeo Lux and Velodyne PUCK), which both use the 905 nm wavelengths, which are used in more than 95 % of currently available LiDARs. The performance was tested and estimated in stabilized fog conditions at Cerema fog chamber facilities. This provides a good basis for repeating the same validation procedure multiple times and ensuring the right development decisions. However, performance of the LiDARs suffers when the weather conditions become adverse and visibility range decreases. A 50 % reduction in target detection performance was observed over the exhaustive tests. Therefore, changing to higher wavelengths (1550 nm) was considered using redesigned “pre-prototype LiDAR”. The preliminary results indicate that there is no reason to not use 1550 nm wavelength, which due to eye safety regulations gives an opportunity to use 20 times more power compared to the traditional 905 nm. In order to clarify the expected benefits, additional feasibility studies are still needed.
关键词： self-driving cars,LiDAR,sensor performance,1550 nm,905 nm,adverse weather,rain,fog
更新于2025-09-04 15:30:14