# Formula Student - Real World AI Presentation

###### tags: `fsfeup`

### Important take aways

**S4.1.2** Half of the Real World AI presentation marks will be awarded for analysis on a minimum of 3 topics such as (but not limited to):
- Other road users including vehicles and pedestrians and other obstacles
- Inclement weather (snow, rain, ice etc)
- Any aspects of connectivity including [V2V](https://www.nhtsa.gov/technology-innovation/vehicle-vehicle-communication) and [V2I](https://blog.rgbsi.com/what-is-v2i-technology) technology
- Functional safety requirements for autonomous vehicles
- The wider philosophy and socioeconomic consequences of a driverless world including ethical and legal issues
- Sensor suite option evaluations

The other half of the presentation marks will be awarded for a detailed analysis of a specific case study. This case study topic for 2023 is **‘Is the world ready for Level 4 autonomous vehicles deployed on public roads?’**

Teams must analyse two or more countries around the world and critically evaluate their legislation, society and technology readiness for driver-out vehicles to be deployed on public roads. Marks will be awarded for analysing the challenges facing the industry and society.

**S4.1.3** The team should focus on the challenges that would be faced for the ADS vehicle to meet SAE Level 4 and can comment on the additional challenges in order to achieve Level 5.

**S4.1.4** Presentations will be evaluated on content, organisation and visual aids as well as the presenters’ delivery and the team’s overall knowledge and response to questions

#### FSAE Levels


---

## Presentation Structure

**1. Introduction: (30 sec max)**

- Present the speakers
- Define the objective of the presentation: evaluating the real world challenges for SAE Level 4 and Level 5 vehicles operating on existing and future road networks.

**2. Challenges for Autonomous Vehicles: (7:30 min; 1:15 per topic)**

- Other road users, including vehicles, pedestrians, and other obstacles
- Sensor suite option evaluations
- Inclement weather (snow, rain, ice etc)
- Any aspects of connectivity including V2V and V2I technology
- Functional safety requirements for autonomous vehicles
- The wider philosophy and socioeconomic consequences of a driverless world including ethical and legal issues

**3. Case Study: "Is the world ready for Level 4 autonomous vehicles deployed on public roads?" (7 min; 2 min per topic)**

- Transition from more general challenges to the analysis of two specific countries regarding their legislation, society, and technology readiness for deploying Level 4 autonomous vehicles. (30 sec max)
- Present countries and why we choose them? (Optional)

- Assess the existing laws and regulations governing autonomous vehicles in each country.
- Analyze the acceptance and preparedness of society for autonomous vehicles, considering factors such as public perception, trust, and concerns.
- Evaluate the technological infrastructure, including road infrastructure, communication networks, and support systems necessary for Level 4 autonomy.

**4. Conclusion: (30 sec max)**

- Identify the key challenges faced by the industry and society in adopting Level 4, 5 autonomous vehicles.

---

## Notes for the presentation

**1. Introduction:**

`Slide - FSFEUP`

Ladies and gentlemen,

`Slide - REAL WORLD AI`

Welcome to the Real World AI event! I'm João, and alongside me is Tomás.

`Slide - Table of contents`

Today, we are thrilled to share our insights into the contemporary challenges encountered by autonomous vehicles and how the industry is actively addressing these issues.

`Slide - Real World Challenges ...`

**2. Challenges for Autonomous Vehicles: (7:30 min; 1:15 per topic)**

### Other road users, including vehicles, pedestrians, and other obstacles.

`Slide - Other Road Users`

Without wasting any time, let's start by exploring the challenges posed by other road users. Specifically, we'll focus on AVs impact on vulnerable road users, including pedestrians, cyclists, motorcyclists, and more.

`Slide - AV Crash Analysis`

To gain a deeper understanding of this topic, we turn to AV crash data analysis. By examining crash data, we can better understand the dynamics and complexities of VRU interactions with AVs. This knowledge serves as a foundation for developing effective safety measures and design strategies.

A recent study analyzed four years of actual AV crash data from California, spanning from 2017 to 2020.

The findings revealed that VRUs were directly and indirectly involved in a subset of AV crashes. In direct crashes, bicyclists were often found at fault, while pedestrians were predominantly involved indirectly.

It is important to note that crashes involving VRUs indirectly occurred more frequently when AVs were in autonomous mode, but the resulting damages were generally minor.

Additionally, the study identified key predictors of VRU-AV-related crashes, including factors such as crosswalks, intersections, traffic signals, and AV movements.

These findings provide valuable insights for AV operators and city planners, to enhance road safety for all road users.

### Sensor suite option evaluations

`Slide - Sensors Analysis`

Regarding the sensor suite options for the development of an ADV, the most commonly used technologies for this purpose are:

- **LiDAR sensors**, which allow for very precise 3D mapping and object detection
- **Radar sensors**, which are very helpful in detecting objects, measuring distances, and providing velocity information
- **Cameras**, used in order to analyze and recognize the detected objects and obtain traffic sign and lane detection information
- **Sonar sensors**, which are used for close-range object detection as well as for parking assistance

### Inclement weather (snow, rain, ice etc)

`Slide - Inclement Weather`

All these sensors may suffer from adverse weather conditions, such as precipitation, fog, or ice.

`Slide - Impact of Inclement Weather`

These conditions present unique obstacles for autonomous systems leading to potential errors in object detection and decision-making. Putting into question the performance and safety of AVs.

Let's consider the impact of rain as an illustrative example. Rain affects the propagation of electromagnetic signals in two distinct ways: Firstly, the absorption of EM energy by water droplets causes attenuation. Secondly, the rain volume backscattering can generate false alarms or mask actual targets in front of the sensor.

A closer examination of the graph reveals a clear relationship between rain rate levels and the visibility of LiDAR~~, a commonly used sensor in AVs~~. As rain intensity increases, the visibility of LiDAR subject to Mie scattering from rain decreases.

`Slide - Overcoming inclement weather`

So, to overcome these problems, autonomous vehicles employ sensor fusion techniques, integrating data from multiple sensors to leverage their respective strengths in different weather conditions. Advanced weather modelling and prediction systems further enable the vehicles to adapt their behaviour accordingly. Not to mention, technological advancements in areas such as AV communication which brings us to our next topic.

### Any aspects of connectivity including V2V and V2I technology

`Slide - Connectivity`

AV connectivity.

`Slide - Connectivity Communications`

Cooperative intelligent transport systems, or ITS, integrate communication technologies that enable vehicles to exchange real-time information and colaborate with other vehicles, infrastructures, and more.

`Slide - Connectivity Technologies`

The wireless technologies used in autonomous vehicles can be differentiated in terms of their transmission range and specific application.

Short-range technologies can assist localisation in dense environments as they do not require the Line of Sight and can easily penetrate in the obstacles.

Medium-range, like DSRC, helps establish a traffic management system which includes highway fleet management, safe overtaking, etc.

Finally long-range have the potential to contribute to the security and privacy areas.
~~, through approaches such as session based authentication and authorisation. They may also benefit in Management and Orchestration and Quality of Service.~~

`Slide - Connectivity Challenges`

But these systems face various challenges. Standardization, privacy, security and interoperability, are crucial sectors that require attention to ensure a successfull implementation of ITS.

One major challenge is the exposure of critical vehicular information in VANETs, which emphasizes the need for robust privacy protection mechanisms. Other important considerations are: the high cost involved in developing and deploying the necessary road infrastructure, and ensuring reliable data transmission between AVs with high mobility and varying location.

### Functional safety requirements for autonomous vehicles
`Slide - Functional Safety Requirements`

Regarding the Functional Safety Requirements, we point out some important aspects of the Hazard Analysis and Risk Assessment process, as well as the redundancy and diversity concepts.

`Slide - Hazard Analysis and Risk Assessment`

Hazard Analysis and Risk Assessment (or HARA) is a crucial process in the development of L4 and L5 ADVs.

The main key steps of the HARA process are:

* **Hazard Identification**: which consists of systematically identifying potential hazards that may arise during the operation of an ADV. Hazards can include both internal and external factors
* **Risk Assessment**: involves evaluating the likelihood of a hazard occurring and the severity of its consequences
* **Risk Mitigation**: where after identifying and assessing risks, measures are put in place to mitigate or reduce them. This may involve design changes and system redundancies
* **Safety Verification and Validation**: where rigorous testing and verification processes are carried out to ensure that the ADV meets safety requirements

`Slide - Redundancy and Diversity`

When developing an ADV that meets SAE L4 or L5 requirements, its important to make use of the **redundancy** and **diversity** concepts:

* **Redundancy**: involves duplicating components/subsystems of an autonomous vehicle to ensure continued functionality in the event of a failure. The idea is to have multiple independent systems performing critical functions, so that if/when one system fails, the redundant system can take over seamlessly
* **Diversity**: refers to the use of different technologies, approaches, or algorithms to perform a specific task. The aim is to reduce the likelihood of failures affecting all systems simultaneously

### The wider philosophy and socioeconomic consequences of a driverless world including ethical and legal issues

`Slide - Ethical Dilemmas and Decision Making`

ADVs may face ethical dilemmas in situations where collision is **unavoidable** and harm to one party is inevitable. This can mean having an ADV decide between causing potential harm to vehicle occupants to avoid hitting VRUs.

Some aspects to consider regarding this topic and during the development of ADV decision-making algorithms are:

* The value of human life
* Trying to minimize harm to all parties involved
* Making sure to always follow the traffic laws and regulations

`Slide - Workforce Disruption`

The significant disruption in the transportation industry that comes as a certain result of the widespread adoption of L4/L5 ADVs is a problem that must be addressed to minimize the socioeconomic impact.

As we can see, according to this study conducted by Erica Groshen, which predicts ADVs' contribution to the unemployment rate in the upcoming years, L4 and L5 vehicles may be responsible for more than 0.12% of the unemployment rate in the next 20 to 25 years.

**3. Case Study: "Is the world ready for Level 4 autonomous vehicles deployed on public roads?" (7 min; 2 min per topic)**

- Transition from more general challenges to the analysis of two specific countries regarding their legislation, society, and technology readiness for deploying Level 4 autonomous vehicles. (30 sec max)
- Present countries and why we choose them? (Optional)

### Introduction

`Slide - Case Study`

The deployment of Level 4 autonomous vehicles on public roads poses significant challenges and requires careful consideration of factors such as legislation, societal acceptance, and technological readiness. This case study focuses on Germany and Japan, two world-leading countries regarding autonomous driving technologies, examining their respective progress in these areas.

### Legislation

**Germany:**
In June 2017, Germany amended its Road Traffic Act to allow drivers to transfer the control of their SAE Level 3 vehicles to an automated driving system and for those vehicles to be used on public roads.

On July 2021, the Autonomous Driving Act entered into force in Germany. This Act allows Level 4 vehicles with autonomous driving capabilities and no driver to circulate in specified operating areas on public roads. A technical supervisor who can deactivate the vehicle from the outside remains necessary.

Deutsche Bahn (DB), the national rail services of Germany, in partnership with Mobileye, ~~a subsidiary of Intel,~~ has unveiled a significant project in the German state of Hesse. The goal is to have over 100 on-demand SAE L4 vehicles operating by the end of the year for public transportation purposes.

**Japan:**
As for Japan, in April 2022 it's Road Traffic Law underwent a revision, with the new regulations taking effect in April of the following year. As per the revision, Level 4 automated driving is now permitted in the country, subject to remote monitoring.

This revision envisions Level 4 automated driving for buses that run unmanned on specific routes mainly in depopulated areas. ~~under the condition of having permission from the Public Safety Commission (Article 75-12, Paragraph 1 of the Road Traffic Act).~~

A notable implementation of Level 4 autonomous driving can be seen in the town of Eiheiji in Fukui Prefecture. The town has received approval to operate a transportation service using three seven-seater electric carts that will traverse paths designated for cyclists and pedestrians.~~These carts will have a single monitoring official, as required by the permits.~~

Looking ahead, the Japanese government aims to expand Level 4 autonomous driving services to at least 40 locations by fiscal year 2025. The focus will be on serving elderly populations in areas where public transportation services have declined.

### Society

`Slide - Individual Acceptance`

Shifting our focus to the topic of societal acceptance of AVs, we have gathered information from surveys conducted in Japan and Germany, as well as data from the KPMG autonomous vehicle readiness index spanning from 2018 to 2020, which revealed interesting patterns in acceptance levels.

Higher AV expectations were linked to higher probabilities of consumer acceptance. However, Germany tended to express a stronger negative sentiment compared to Japan throughout the years.

`Slide - Society Key Components`

When examining the key drivers of such components, Germany placed greater emphasis factors like happiness and social issues, while convenience held more significance in Japan. These differences indicated the need for tailored marketing strategies for promoters of AVs in each country.

Surprisingly, as knowledge of AVs increases, acceptance decreases, suggesting the influence of negative public messages surrounding AVs.

`Slide - KPMG AVRI`

Turning our attention to the KPMG autonomous vehicle readiness index, we observe fluctuations over the years, with Germany experiencing a decline in its consumer acceptance ranking. This can be attributed to changes in comparison measures, including the removal of a survey-based measure and the addition of new indicators ~~such as digital skills and individual readiness,~~ where countries like Japan excel more.

### Technological Readiness

Both Germany and Japan possess a robust technological ecosystem that paves the way for the widespread adoption of L4 and L5 ADVs on public roads.

**Germany:**
German automakers like Audi, BMW, and Mercedes, have been at the forefront, investing heavily in autonomous driving technology. This commitment to innovation propels advancements in the field.

Furthermore, the government actively supports and encourages the development and testing of ADVs. A notable initiative is the "Digital Test Field Autobahn", established in September 2015 on the A9 in Bavaria. Its objective is to test modern and future-oriented technologies, such as ADVs, in real traffic scenarios.

**Japan:**
Regarding Japan, the country's ageing population has created a strong need for safer and more comfortable transportation options, making autonomous vehicles a priority.

One of Japan's notable strengths lies in its patent leadership. For many years, Japan has dominated the AV-related patent market, showcasing its commitment to research and development.

In terms of infrastructure, Japan possesses excellent road quality and extensive mobile coverage. However, the country does face challenges with a significant number of tunnels and narrow urban streets.

To foster the development and testing of autonomous vehicles, Japan has also established a dedicated Automated Driving Test Center enabling a wide range of testing scenarios. ~~including various weather conditions, communication systems in V2X areas, and complex urban intersections.~~

In conclusion, all these factors position Japan and Germany as key players in shaping the future of autonomous mobility.

### Conclusion (30 sec max)

Overall, in response to the question of whether the world is ready for Level 4 autonomous vehicles deployed on public roads, our answer is no. We have witnessed, particularly in Japan and Germany, progress that suggests we are not far from realizing this reality. While challenges remain, through continued research, industry collaboration, and many other solutions, we will overcome these obstacles and pave the way for the widespread acceptance and integration of autonomous vehicles.

---

## Notes on society

Surveys were conducted in Japan (January 2017) and in Germany (December 2018) evenly distributed by age, gender, and region. These were analysed using Principal Component Analysis and then with an Ordered Logit Model.

National-level cultural differences:
- **Japan:** Highly masculine, long-term orientation, preference for certainty, hierarchical, and collective.
- **Germany:** Fewer masculinity and power distance, but similar to Japan in terms of long-term orientation and favouring restraint.

The following results were drawn from the responses:

1. Acceptance Levels:
- **Japan:** The acceptance of partial AVs is significantly higher than fully AVs. The acceptance of partial AVs is broadly positive, while the acceptance of fully AVs is also positive but to a lesser extent.
- **Germany:** The acceptance of partial AVs is significantly higher than fully AVs. The acceptance is lower, with a strong negative sentiment towards both levels of automation.
Germans are more sensitive to privacy issues than the Japanese.

2. Components of Acceptance:
- Principal Component Analysis (PCA) identified five components that summarize the variables related to AV acceptance: AV Expectations, AV Ignorance, Driver Exposure, AV Fear, and Passenger Exposure.
- The importance of convenience in AV expectations is higher in Japan compared to Germany, where happiness and social issues are more significant.
- AV ignorance shows a surprising result that acceptance decreases as knowledge of AVs increases in all three countries. The negative perception of AVs is dominant in public messages, highlighting the need for the industry to change this perception.
- Driver exposure has different influential factors in each country. In Japan, shared trips and minutes are predominant. In Germany, frequent drivers who share their vehicles show the most acceptance.
- AV fear is most influenced by pedestrians, but partially AVs evoke slightly more fear than fully AVs.
- Passenger exposure shows different patterns between Japan and the UK/Germany, suggesting potential value in organizing events to expose non-drivers to AVs in a protected environment.

3. Factors Predicting Acceptance:
- Gender, age, region, car ownership, car license, and the identified components were used as independent variables to predict individual acceptance.
- The models showed that attitudes and components had a stronger influence on acceptance than socio-economic factors.
- Socio-economic variables were not found to be statistically significant predictors of AV acceptance.
- The UK showed stronger effects for AV expectations, AV ignorance, AV fear, and driver exposure in both partial and full AV models compared to Germany and Japan.

4. Cultural Differences:
- The higher masculinity score for Japan could explain why gender was significant there but not in the UK or Germany (Male respondents are more likely to accept fully AVs than females).
- Other dimensions (region, car ownership, car license) did not show clear patterns concerning AV acceptance.

KPMG:
- 2018
- Population living in test areas (Japan bigger than Germany)
- Consumer survey data on AV acceptance (Germany bigger)
- KPMG Change Readiness people and civil society technology use (Germany bigger)
- WEF GCI technology readiness (Germany bigger very close)
- 2019
- Population living in test areas (Japan bigger)
- Consumer survey data on AV acceptance (Japan bigger)
- KPMG Change Readiness people and civil society technology use (Germany much bigger)
- WEF GCI technology readiness (Germany bigger very close)
- \+ Ridesharing market penetration (Germany bigger)
- 2020
- Population living in test areas (Japan bigger)
- KPMG Change Readiness people and civil society technology use (Germany much bigger)
- \+ Consumer ICT adoption (Japan bigger)
- \+ Digital skills (Germany bigger)
- \+ Individual readiness (Japan bigger)
- \+ Ridesharing market penetration (Germany bigger very close)

1. Acceptance Levels:
- Japan has higher acceptance of partial AVs compared to fully AVs, while both levels of automation are viewed positively.
- Germany has lower acceptance of both partial and fully AVs, with a strong negative sentiment towards automation. Germans are more sensitive to privacy issues.

2. Components of Acceptance:
- AV Expectations: Convenience is more important in Japan, while happiness and social issues are more significant in Germany.
- AV Ignorance: Acceptance decreases as knowledge of AVs increases in both countries. Negative perception of AVs is dominant in public messages, suggesting a need for industry intervention.
- AV Fear: Pedestrians have the most influence on AV fear, with slightly more fear associated with partial AVs compared to fully AVs.

3. Factors Predicting Acceptance:
- Attitudes and components have a stronger influence on acceptance than socio-economic factors.
- Gender is a significant predictor in Japan but not in the UK or Germany. Male respondents in Japan are more likely to accept fully AVs than females.
- Other dimensions such as region, car ownership, and car license do not show clear patterns concerning AV acceptance.

Based on these points, it can be concluded that Japan generally has higher acceptance of AVs, especially in the case of partial automation, compared to Germany. Cultural factors, including masculinity and attitudes towards convenience and happiness/social issues, play a role in shaping AV acceptance in each country. Privacy concerns are more pronounced in Germany. Additionally, the level of digital skills, individual readiness, and online ride-hailing market penetration differ between the two countries, potentially influencing AV acceptance levels.