A survey of deep learning techniques for autonomous driving

Grigorescu, Sorin Mihai; Trasnea, Bogdan; Cocias, Tiberiu T.; Macesanu, Gigel

doi:10.1002/rob.21918

Cited by 1,115 publications

(581 citation statements)

References 134 publications

Supporting

Mentioning

495

Contrasting

Unclassified

Order By: Relevance

“…Another interesting aspect for RL drivers is learning to overtake other cars, which can be a particularly challenging task, depending on the shape of the road section in which the cars are placed [19], but also depending on the vehicles size, as in [20], where a RL driver learns to control a truck-trailer vehicle in an highway with other regular cars. The authors of [21,22] provided extensive classifications of the AI state-of-the-art techniques employed in autonomous driving, together with the degrees of automation that are possible for self-driving cars.…”

Section: Related Workmentioning

confidence: 99%

On the Impact of the Rules on Autonomous Drive Learning

et al. 2020

View full text Add to dashboard Cite

Autonomous vehicles raise many ethical and moral issues that are not easy to deal with and that, if not addressed correctly, might be an obstacle to the advent of such a technological revolution. These issues are critical because autonomous vehicles will interact with human road users in new ways and current traffic rules might not be suitable for the resulting environment. We consider the problem of learning optimal behavior for autonomous vehicles using Reinforcement Learning in a simple road graph environment. In particular, we investigate the impact of traffic rules on the learned behaviors and consider a scenario where drivers are punished when they are not compliant with the rules, i.e., a scenario in which violation of traffic rules cannot be fully prevented. We performed an extensive experimental campaign, in a simulated environment, in which drivers were trained with and without rules, and assessed the learned behaviors in terms of efficiency and safety. The results show that drivers trained with rules enforcement are willing to reduce their efficiency in exchange for being compliant to the rules, thus leading to higher overall safety.

show abstract

Section: Related Workmentioning

confidence: 99%

On the Impact of the Rules on Autonomous Drive Learning

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Recently, deep learning (DL) methods have proven themselves worthy of consideration in microscopy image analysis tools as they have also been successfully applied in a wider range of applications including but not limited to face detection (Taigman et al, 2014, Sun et al 2014, Schroff et al 2015, self-driving cars (Badrinarayanan et al, 2017, Redmon et al, 2016, Grigorescu et al, 2019 and speech recognition (Hinton et al, 2012). Caicedo et al (Caicedo et al, 2019) and others (Hollandi et al 2019, Moshkov et al 2019 proved that single cell detection and segmentation accuracy can be significantly improved utilizing DL networks.…”

Section: Introductionmentioning

confidence: 99%

“…As robustly and accurately as they may perform, these networks rely on sufficient data, both in amount and quality, which tends to be the bottleneck of their applicability in certain cases such as single-cell detection. While in more industrial applications (see (Grigorescu et al, 2019)for an overview of autonomous driving) a large amount of training data can be collected relatively easily: see the cityscapes dataset (Cordts et al, 2016) (available at https://www.cityscapes-dataset.com/ ) of traffic video frames using a car and camera to record and potentially non-expert individuals to label the objects, clinical data is considerably more difficult, due to ethical constraints, and expensive to gather as expert annotation is required. Datasets available in the public domain such as BBBC (Ljosa et al, 2012) at https://data.broadinstitute.org/bbbc/ , TNBC (Naylor et al, 2017(Naylor et al, , 2019 or TCGA (Cancer Genome Atlas Research Network, 2008;Kumar et al, 2017) and detection challenges including ISBI (Coelho et al, 2009), Kaggle ( https://www.kaggle.com/ ), ImageNet (Russakovsky et al, 2015) etc.…”

Section: Introductionmentioning

confidence: 99%

AnnotatorJ: an ImageJ plugin to ease hand-annotation of cellular compartments

Hollandi

Horváth

2020

Preprint

View full text Add to dashboard Cite

When single-cell identification and deep learning meet, AnnotatorJ arises. Cellular analysis quality depends on accurate and reliable detection and segmentation of cells so that the subsequent steps of analyses e.g. expression measurements may be carried out precisely and without bias. Deep learning has recently become a popular way of segmenting cells, performing unimaginably better than conventional methods. However, such deep learning applications may be trained on a large amount of annotated data to be able to match the highest expectations. High-quality annotations are unfortunately expensive as they require field experts to create them, and often cannot be shared outside the lab due to medical regulations.We propose AnnotatorJ, an ImageJ plugin for the semi-automatic annotation of cells (or generally, objects of interest) on (not only) microscopy images that helps find the true contour of individual objects by applying U-Net pre-segmentation. The manual labour of hand-annotating cells can be significantly accelerated by using our tool. Thus, it enables users to create such datasets that could potentially increase the accuracy of state-of-the-art solutions, deep learning or otherwise, when used as training data.

show abstract

“…Among the new technological trends arisen in the last decade, Autonomous Driving has gained a lot of attention, with significant effort and resources invested by both academia and enterprises. The breakthroughs in self-driving cars have been made possible by the emergence of novel algorithms and real-time computing systems in the field of Artificial Intelligence (AI), deep learning (DL), Internet-of-Things (IoT) and Cloud computing [ 1 ].…”

Section: Introductionmentioning

confidence: 99%

Cloud2Edge Elastic AI Framework for Prototyping and Deployment of AI Inference Engines in Autonomous Vehicles

Grigorescu

Cocias

Trasnea

et al. 2020

Sensors

Self Cite

View full text Add to dashboard Cite

Self-driving cars and autonomous vehicles are revolutionizing the automotive sector, shaping the future of mobility altogether. Although the integration of novel technologies such as Artificial Intelligence (AI) and Cloud/Edge computing provides golden opportunities to improve autonomous driving applications, there is the need to modernize accordingly the whole prototyping and deployment cycle of AI components. This paper proposes a novel framework for developing so-called AI Inference Engines for autonomous driving applications based on deep learning modules, where training tasks are deployed elastically over both Cloud and Edge resources, with the purpose of reducing the required network bandwidth, as well as mitigating privacy issues. Based on our proposed data driven V-Model, we introduce a simple yet elegant solution for the AI components development cycle, where prototyping takes place in the cloud according to the Software-in-the-Loop (SiL) paradigm, while deployment and evaluation on the target ECUs (Electronic Control Units) is performed as Hardware-in-the-Loop (HiL) testing. The effectiveness of the proposed framework is demonstrated using two real-world use-cases of AI inference engines for autonomous vehicles, that is environment perception and most probable path prediction.

show abstract

A survey of deep learning techniques for autonomous driving

Cited by 1,115 publications

References 134 publications

On the Impact of the Rules on Autonomous Drive Learning

On the Impact of the Rules on Autonomous Drive Learning

AnnotatorJ: an ImageJ plugin to ease hand-annotation of cellular compartments

Cloud2Edge Elastic AI Framework for Prototyping and Deployment of AI Inference Engines in Autonomous Vehicles

Contact Info

Product

Resources

About