Why We should be Careful when Using Self-driving Vehicle Features!

Schwarz, DI Manuel; Rolland, Luc; Johnston, James

doi:10.1109/ice/itmc52061.2021.9570269

Cited by 2 publications

(4 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On May 7, 2016, while the Tesla autopilot mode was active, a vehicle driver died in a fatal crash when the system did not detect the motorway crossing tractor-trailer correctly [1]. In 2018, another fatal accident occurred during one of Uber's self-driving tests during the night, [3]. The vehicle could not sense and avoid a crossing pedestrian, and the driver was not actively looking at the street since she was engaged on her mobile phone.…”

Section: B Contextmentioning

confidence: 99%

“…The newly available driving freedom brings many benefits and some critical safety drawbacks. The various activities which can be performed during a trip increase distraction, which significantly increases the stopping distance when a problem occurs [3]. As long as the industry and consumers are in a transition phase, special caution is required when utilizing new systems.…”

Section: B Contextmentioning

confidence: 99%

See 1 more Smart Citation

Validation of vehicles with automated features through the detection of behavioral differences between humans and machines

Schwarz

Rolland

Johnston

2022

2022 IEEE 28th International Conference on Engineering, Technology and Innovation (ICE/ITMC) &Amp; 31st International Associati

Self Cite

View full text Add to dashboard Cite

The development of automated vehicles has not achieved the desired success over the past decade, despite numerous technological and methodological successes. Currently, no production-ready autonomous vehicle on the market can operate without a human driver in a wide range of regions and environmental conditions. Due to the high complexity of the systems and the varied traffic scenarios, development, and validation are challenging and timeconsuming. With the aid of simulation technology, good validation in a scalable form can be performed during development. However, the current standards for traffic safety validation require a final validation under real conditions. In this work, a framework called Shadow System (SS) is presented, which independently captures the control commands of the human driver and the vehicle. This unique approach enables the detection of problems in automated systems by identifying behavioral differences. Up to a defined safety range allows the human driver and the vehicle to execute control commands without affecting each other. This makes it possible, for example, to determine how a human driver steers the vehicle compared to the automated system in a given traffic situation. The developed method enables safe use under real conditions, as control can be taken over directly by the human driver at any time. Scenarios identified as critical will provide information about the safety of the system during the validation process.

show abstract

Section: B Contextmentioning

confidence: 99%

Section: B Contextmentioning

confidence: 99%

Validation of vehicles with automated features through the detection of behavioral differences between humans and machines

Schwarz

Rolland

Johnston

2022

2022 IEEE 28th International Conference on Engineering, Technology and Innovation (ICE/ITMC) &Amp; 31st International Associati

Self Cite

View full text Add to dashboard Cite

show abstract

“…The growth of Advanced Driver Assistance Systems (ADASs) has the potential impact to improve the global industries engaged in transportation by generating cheaper, faster, and safer architecture [1][2][3][4][5]. The ADAS has been a significant automotive industry focus in recent years.…”

Section: Introductionmentioning

confidence: 99%

“…In other words, a key challenge to building ADASs is to safely react and accurately perceive the diversity of traffic-risk prediction in real-world applications. Nowadays, the traffic-risk incident is attractive research due to the limitation of data resources and distributions [5,10,11]. For guaranteeing a safe driving strategy, the driving scenario obeys all incident cases with a long-tailed distribution, such that a minimal number of common situations makes up the vast majority of what a driver encounters and a virtually infinite number of rare scenarios [2].…”

Section: Introductionmentioning

confidence: 99%

An End-to-End Online Traffic-Risk Incident Prediction in First-Person Dash Camera Videos

Pradana

2023

BDCC

View full text Add to dashboard Cite

Predicting traffic risk incidents in first-person helps to ensure a safety reaction can occur before the incident happens for a wide range of driving scenarios and conditions. One challenge to building advanced driver assistance systems is to create an early warning system for the driver to react safely and accurately while perceiving the diversity of traffic-risk predictions in real-world applications. In this paper, we aim to bridge the gap by investigating two key research questions regarding the driver’s current status of driving through online videos and the types of other moving objects that lead to dangerous situations. To address these problems, we proposed an end-to-end two-stage architecture: in the first stage, unsupervised learning is applied to collect all suspicious events on actual driving; in the second stage, supervised learning is used to classify all suspicious event results from the first stage to a common event type. To enrich the classification type, the metadata from the result of the first stage is sent to the second stage to handle the data limitation while training our classification model. Through the online situation, our method runs 9.60 fps on average with 1.44 fps on standard deviation. Our quantitative evaluation shows that our method reaches 81.87% and 73.43% for the average F1-score on labeled data of CST-S3D and real driving datasets, respectively. Furthermore, the proposed method has the potential to assist distribution companies in evaluating the driving performance of their driver by automatically monitoring near-miss events and analyzing driving patterns for training programs to reduce future accidents.

show abstract

Why We should be Careful when Using Self-driving Vehicle Features!

Cited by 2 publications

References 10 publications

Validation of vehicles with automated features through the detection of behavioral differences between humans and machines

Validation of vehicles with automated features through the detection of behavioral differences between humans and machines

An End-to-End Online Traffic-Risk Incident Prediction in First-Person Dash Camera Videos

Contact Info

Product

Resources

About