This study focuses on the industry practices for challenging autonomous driving systems with critical scenarios. The complexity of testing these systems for safety and reliability is highlighted, particularly in identifying relevant test scenarios that may expose hazards or risks to both autonomous vehicles and other road users. Through 10 interviews with 13 participants from 7 companies in Sweden involved in autonomous driving, thematic modeling was used to analyze the data. The results revealed limited joint efforts in the industry to explore different approaches and tools for critical scenario identification. In the related work section, various studies on critical scenario identification are discussed, emphasizing the importance of understanding what is considered critical and how to select these scenarios due to the infinite number of possible scenarios in operational environments. Different approaches are proposed, including using experts, naturalistic driving data, or guided search. A taxonomy based on a systematic literature review categorizes these approaches into four categories: exploration without parameter trajectory, exploration with parameter trajectories, induced learning from existing data sources, and deductive learning using domain expertise. The methodology used for conducting interviews is outlined in this article along with a detailed list of interview participants and their respective lengths and formats. The structured approach taken during interviews aimed at setting a basis/context for progressive discussions. Overall, this study contributes to exploring industry practices and challenges related to critical scenario identification and testing for autonomous driving systems. Recommendations include combining different approaches, collaborating among stakeholders, and continuously learning in this field to enhance safety and reliability testing practices within the industry.
- Study focuses on industry practices for challenging autonomous driving systems with critical scenarios
- Complexity of testing autonomous driving systems for safety and reliability is highlighted
- Limited joint efforts in the industry to explore different approaches and tools for critical scenario identification
- Importance of understanding what is considered critical and how to select these scenarios due to infinite possible scenarios
- Different approaches proposed for critical scenario identification: experts, naturalistic driving data, guided search
- Taxonomy categorizes approaches into four categories: exploration without parameter trajectory, exploration with parameter trajectories, induced learning from existing data sources, deductive learning using domain expertise
- Methodology used for conducting interviews outlined in the article
- Recommendations include combining different approaches, collaborating among stakeholders, continuously learning to enhance safety and reliability testing practices within the industry
Summary- The study looks at how to test self-driving cars in tough situations.
- Testing these cars for safety and reliability is very hard.
- Not many companies work together to find the best ways to identify tough scenarios.
- It's important to know what situations are critical and how to pick them out from many possibilities.
- There are different ways to find tough scenarios, like using experts or real driving data.
Definitions- Autonomous driving systems: Cars that can drive by themselves without a human controlling them.
- Scenarios: Different situations or events that can happen.
- Critical: Very important or serious.
- Identification: Finding or recognizing something.
- Taxonomy: A way of organizing things into groups based on their similarities.
Introduction
The development of autonomous driving systems (ADS) has been a major focus in the automotive industry in recent years. These systems have the potential to revolutionize transportation by increasing safety, efficiency, and accessibility on our roads. However, with this new technology comes the challenge of ensuring its safety and reliability through rigorous testing.
In order to properly test ADS, it is crucial to identify critical scenarios that may expose hazards or risks to both autonomous vehicles and other road users. This is a complex task due to the infinite number of possible scenarios that can occur in real-world environments. In this research paper, titled "Industry Practices for Challenging Autonomous Driving Systems with Critical Scenarios," a study was conducted to explore how companies in Sweden are approaching this challenge.
Methodology
The researchers used thematic modeling to analyze data collected from 10 interviews with 13 participants from 7 companies involved in autonomous driving in Sweden. The methodology used for conducting these interviews is outlined in detail within the paper, including information about the interview participants and their respective lengths and formats.
A structured approach was taken during the interviews, aiming at setting a basis/context for progressive discussions. This allowed for a deeper understanding of industry practices and challenges related to critical scenario identification and testing for ADS.
Results
The results revealed limited joint efforts among companies in Sweden when it comes to exploring different approaches and tools for critical scenario identification. This highlights the need for collaboration among stakeholders in order to enhance safety and reliability testing practices within the industry.
In addition, various studies on critical scenario identification were discussed in relation to their approaches and methodologies. It was found that there are four main categories of approaches: exploration without parameter trajectory, exploration with parameter trajectories, induced learning from existing data sources, and deductive learning using domain expertise.
Exploration without Parameter Trajectory
This approach involves exploring scenarios without any predefined parameters or trajectories. It relies on the expertise and intuition of the individuals conducting the exploration.
Exploration with Parameter Trajectories
In this approach, scenarios are explored based on predefined parameters and trajectories. This allows for a more systematic and controlled exploration process.
Induced Learning from Existing Data Sources
This approach involves using existing data sources, such as naturalistic driving data, to identify critical scenarios. By analyzing real-world driving data, potential hazards can be identified and used to create test scenarios.
Deductive Learning Using Domain Expertise
The final category of approaches involves using domain expertise to deductively identify critical scenarios. This may involve experts in specific fields, such as human factors or traffic engineering, who can provide insight into potential hazards and risks in different driving situations.
Discussion and Recommendations
Based on the results of this study, it is clear that there is no one-size-fits-all approach when it comes to identifying critical scenarios for testing ADS. Instead, a combination of different approaches may be necessary in order to fully capture all potential hazards and risks.
Collaboration among stakeholders is also crucial in this field. By sharing knowledge and resources, companies can work together towards enhancing safety and reliability testing practices for ADS.
Continuous learning is also important in order to keep up with advancements in technology and new challenges that may arise. As autonomous driving continues to evolve, so too must our methods for testing its safety and reliability.
In conclusion, this research paper provides valuable insights into industry practices for challenging autonomous driving systems with critical scenarios. By understanding these practices and challenges, we can work towards improving safety standards for this emerging technology.
