Requirements on driving dynamics in autonomous driving with regard to motion and comfort

Burkhard, Georg; Vos, Steve De; Munzinger, Nathan; Enders, Erik; Schramm, Dieter

doi:10.1007/978-3-658-21194-3_53

Cited by 17 publications

(6 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Safety requirement violations are by far the most critical requirements, as a misbehaviour in the steering component may cause a crash of the vehicle with potential casualties. However, in general, a SDC might violate also other driving requirements, e.g., related to ride comfort [11], such as excessive derivative of the steering angle, unstable movement around the centerline, or excessive deceleration.…”

Section: Misbehaviour Of Autonomous Vehiclesmentioning

confidence: 99%

Misbehaviour prediction for autonomous driving systems

Stocco

Weiß

Calzana

et al. 2020

Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering

120

131

View full text Add to dashboard Cite

Deep Neural Networks (DNNs) are the core component of modern autonomous driving systems. To date, it is still unrealistic that a DNN will generalize correctly to all driving conditions. Current testing techniques consist of offline solutions that identify adversarial or corner cases for improving the training phase. In this paper, we address the problem of estimating the confidence of DNNs in response to unexpected execution contexts with the purpose of predicting potential safety-critical misbehaviours and enabling online healing of DNN-based vehicles. Our approach SelfOracle is based on a novel concept of self-assessment oracle, which monitors the DNN confidence at runtime, to predict unsupported driving scenarios in advance. SelfOracle uses autoencoderand time series-based anomaly detection to reconstruct the driving scenarios seen by the car, and to determine the confidence boundary between normal and unsupported conditions. In our empirical assessment, we evaluated the effectiveness of different variants of SelfOracle at predicting injected anomalous driving contexts, using DNN models and simulation environment from Udacity. Results show that, overall, SelfOracle can predict 77% misbehaviours, up to six seconds in advance, outperforming the online input validation approach of DeepRoad. CCS CONCEPTS • Software and its engineering → Software testing and debugging.

show abstract

Section: Misbehaviour Of Autonomous Vehiclesmentioning

confidence: 99%

Misbehaviour prediction for autonomous driving systems

Stocco

Weiß

Calzana

et al. 2020

Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering

120

131

View full text Add to dashboard Cite

show abstract

“…Safety requirement violations are by far the most critical requirements, as a misbehaviour in the steering component may cause a crash of the vehicle with potential casualties. However, in general, a SDC might violate also other driving requirements, e.g., related to ride comfort [8], such as excessive derivative of the steering angle, unstable movement around the centerline, or excessive deceleration. All these could also be considered as misbehaviours by our definition.…”

Section: Problem Formulationmentioning

confidence: 99%

Misbehaviour Prediction for Autonomous Driving Systems

Stocco¹,

Weiss²,

Calzana³

et al. 2019

Preprint

View full text Add to dashboard Cite

Deep Neural Networks (DNNs) are the core component of modern autonomous driving systems. To date, it is still unrealistic that a DNN will generalize correctly in all driving conditions. Current testing techniques consist of offline solutions that identify adversarial or corner cases for improving the training phase, and little has been done for enabling online healing of DNN-based vehicles.In this paper, we address the problem of estimating the confidence of DNNs in response to unexpected execution contexts with the purpose of predicting potential safety-critical misbehaviours such as out of bound episodes or collisions. Our approach SelfOracle is based on a novel concept of self-assessment oracle, which monitors the DNN confidence at runtime, to predict unsupported driving scenarios in advance. SelfOracle uses autoencoder and time-series-based anomaly detection to reconstruct the driving scenarios seen by the car, and determine the confidence boundary of normal/unsupported conditions.In our empirical assessment, we evaluated the effectiveness of different variants of SelfOracle at predicting injected anomalous driving contexts, using DNN models and simulation environment from Udacity. Results show that, overall, SelfOracle can predict 77% misbehaviours, up to 6 seconds in advance, outperforming the online input validation approach of DeepRoad by a factor almost equal to 3.

show abstract

“…Passengers of HAVs are allowed to divert their attention to other activities like reading or working (SAE International, 2021). However, visually distracted, passengers are more susceptible to centripetal accelerations and tilt their heads less consistently compared to attentive passengers who are looking ahead (Burkhard et al, 2018). This phenomenon is a result of the unawareness of the upcoming vehicle manoeuvres when passengers' visual attention is diverted from the road ahead and may apply to longitudinal accelerations as well.…”

Section: Introductionmentioning

confidence: 99%

“…Acceleration represents the rate of change of velocity with respect to time, while jerk is the rate at which the vehicle's acceleration changes over time. These forces cause movements of the passenger's head and body and can influence the perception of comfort and well-being, particularly when the passenger is engaged in a non-driving related activity (NDRA) such as reading (Burkhard et al, 2018). Very high jerk can be similar to the experience of a 'kick '-movement (de Winkel et al, 2023).…”

Section: Introductionmentioning

confidence: 99%

User Evaluation of Comfortable Deceleration Profiles for Highly Automated Driving: Findings from a Test Track Study

Horn,

Rossner,

Madigan

et al. 2023

Preprint

View full text Add to dashboard Cite

As automated vehicles advance and become more widespread, it is increasingly important to ensure optimal driving comfort for passengers. Recent research has focused on developing driving styles for automated vehicles that are perceived to be most comfortable. However, there is still little understanding if and how possible driving styles need to be adjusted for specific traffic scenarios. In this study, a total of 36 participants experienced three different deceleration profiles (a linear deceleration profile ‘One-Step’, and two versions of stepwise deceleration profiles ‘Two-Step V1 and V2’) across different driving scenarios (deceleration before curves, speed-limit, and stop sign). Deceleration profiles were rated by participants and the impact of non-driving related activities on driving comfort was investigated. Results show a positive rating for all deceleration profiles in terms of comfort. For decelerations into a stillstand at a Stop Sign, participants seem to prefer the One-Step approach which decelerates continuously. However, participants ranked the smoother Two-Step V1 approach more frequently as their personal favourite than the One-Step profile and the more dynamic Two-Step V2 profile. The visual distraction of the passenger through a non-driving activity had no impact on passenger comfort and preferences for the profiles and scenarios tested within this study. Nonetheless, participants reported perceiving a lower intensity of longitudinal vehicle movements when visually distracted during the drive. The results of the study provide recommendations for the design and implementation of deceleration profiles perceived to be comfortable.

show abstract

Requirements on driving dynamics in autonomous driving with regard to motion and comfort

Cited by 17 publications

References 0 publications

Misbehaviour prediction for autonomous driving systems

Misbehaviour prediction for autonomous driving systems

Misbehaviour Prediction for Autonomous Driving Systems

User Evaluation of Comfortable Deceleration Profiles for Highly Automated Driving: Findings from a Test Track Study

Contact Info

Product

Resources

About