More than a century ago, the invention of automobiles revolutionized people’s lives, and the sector is now preparing for its next big leap, which is autonomous driving. In today’s world, where everything is getting smart, why wouldn’t your cars be smart too? Well, they are.
Technological advancement in the automotive sector has led to the rise of self-driving vehicles.
The combination of powerful hardware and intelligent software is ushering in a new era of self-driving cars that will take people to their destinations comfortably and safely without requiring any human intervention.
This autonomous driving landscape is evolving at a rapid pace, with the number of automated vehicles shipped each year projected to grow at a CAGR of 41% between 2024 and 2030.
The growing integration of automated vehicles into our daily lives is expected to reduce traffic congestion, enhance accessibility, and increase safety. These vehicles further allow drivers to engage in tasks that are not related to driving, such as using phones, watching multimedia, working, or simply relaxing while in transit.
Not all automated vehicles can achieve that, though. There are levels to automated vehicles, which are as follows;
Level 0 – At this level, there’s no driving automation. It is completely manually controlled. This level of vehicles is what we see mostly on the roads.
Level 1 – Now, one step above comes the lowest level of automation, where driver assistance is provided through a single automated system like steering or adaptive cruise control.
Level 2 – This partial driving automation level is advanced driver assistance systems (ADAS). Here, the vehicle can control the steering and its speed, but a human still sits in the driver’s seat and can take control of the vehicle at any time. We are already seeing this level of vehicles around us with Tesla (TSLA +0.59%) Autopilot and General Motors‘ (GM +0.92%) Cadillac Super Cruise systems.
Level 3 – In this conditional automation level, the vehicles have the capabilities to detect their surroundings and, based on that, can make informed decisions. However, the human driver still has to maintain alertness and be ready to take control if the system can’t execute the task. Prime examples of Level 3 include the Mercedes-Benz S-Class, which comes with capabilities like autonomous highway cruising and lane-keeping, and the Honda (HMC -2.82%) Legend, which comes with hands-free driving in specific scenarios.
Level 4 – This next-level jump takes us to high-automation vehicles, which do not require human assistance, in most cases, of course. However, humans still have the option to override manually. Level 4 vehicles can self-drive but legally can only do so in limited areas. These vehicles are already in development with Google‘s Waymo One and Baidu’s Apollo Go.
Level 5 – Now, this level requires no drivers at all. At this stage, autonomous vehicles reach full driving automation, and they are free from geofencing and, as such, able to do all that an experienced human driver can do and go anywhere. From Tesla, Amazon (AMZN +2.38%), and Honda to Mercedes, several major automakers all around the world are testing fully autonomous cars. They, however, are yet to be made available to the general public.
While it is currently unknown just when fully automated vehicles (SAE Level 5) will gain widespread adoption, some studies predict market readiness by the end of this decade.
With that, it is critical to build user trust for the successful deployment and acceptance of these vehicles. Currently, limited passenger trust is obstructing the adoption.
So, to help make self-driven vehicles passenger-friendly, researchers from the Gwangju Institute of Science and Technology (GIST), South Korea, wrote a paper discussing strategies for the same. This includes providing explanations to passengers.
The thing is, poorly designed explanations can adversely affect the passenger experience. Hence, explanations must convey information with sufficient intelligibility under rapidly changing road environments.
Previous studies have explored various explanation presentation methods to enhance passenger experience while reducing anxiety and cognitive load. However, the optimal timing for explanations and the actual passenger demand have yet to be widely explored, especially in real environments.
The GIST researchers investigated the process of providing explanations in a timely manner in order to enhance the sense of safety of passengers and their confidence in automated vehicles.
Accelerating Autonomous Vehicles Adoption
To help automated vehicles live up to their promise of improving urban mobility, passenger trust must be achieved, for which timely, passenger-specific explanations need to be provided for self-driving vehicle decisions.
For these explanations to be effective, they need to be understandable, informative, and concise. That will foster trust among passengers by providing them with an increasing sense of control and reducing negative experiences.
While explainable artificial intelligence (XAI) methods already exist, they are primarily for developers and regulators. With their focus on high-risk scenarios or too detailed explanations, they aren’t really suitable for passengers.
This highlights the need for XAI models that specifically focus on passengers, understanding the type of information required and when it is needed in real-world driving scenarios.
A primary obstacle to developing passenger-centric explainable XAI models, the study noted, is the lack of datasets that account for passenger contexts.
In response, a team of GIST researchers led by SeungJun Kim, the Professor and Director of the Human-Centered Intelligent Systems Lab at GIST, introduced TimelyTale to address the lack of a passenger-centric approach using sensor data for timely and context-relevant explanations.
TimelyTale is a novel multimodal dataset designed to capture real-world driving scenarios and offer in-vehicle explanations to improve the trust and confidence of passengers in automated vehicles.
“Our research shifts the focus of XAI in autonomous driving from developers to passengers. We have developed an approach for gathering passenger’s actual demand for in-vehicle explanations and methods to generate timely, situation-relevant explanations for passengers.”
– Professor Kim
The study authors were awarded ‘Distinguished Paper Award’ for their study titled ‘What and When to Explain?: On-road Evaluation of Explanations in Highly Automated Vehicles.’
To begin, the researchers first looked into the effect of different types of visual explanations— including attention, perception, and a combination of both—as well as their timing on passenger experience under real driving conditions, utilizing augmented reality.
The vehicle’s perception state was found to improve trust, situational awareness, and perceived safety without overwhelming the passengers. Additionally, the researchers found traffic risk probability to be the most effective factor in deciding when explanations should be delivered, which also helped them understand when passengers feel overloaded with information.
Based on these findings, GSIT researchers, in collaboration with MIT, developed the TimelyTale dataset.
For this approach, researchers used data from the external environment (exteroceptive) like sounds and sights, proprioceptive data, which is about the body’s positions and movements, and (interoceptive) data about the passenger’s state, i.e., their body’s sensations like pain, breathing, and heart rate.
To gather all this data from passengers, the researchers used a variety of sensors in naturalistic driving scenarios to predict their explanation demands. The devices used included GPS, 3D LiDAR, OBD-II, IMUs, and stereo cameras for exteroceptive and proprioceptive data, while LiDAR camera, depth camera, thermal imaging, E4 wristband, and seat pressure sensors were used to capture interoceptive data.
Notably, the researchers also incorporated the concept of in-vehicle interruptibility to find the right moments for explanations. Interruptibility is the shift in the passenger’s focus from non-driving related tasks (NDRTS) to driving-related information.
Unlike manually driven vehicles, where drivers can’t be distracted, in automated vehicles, passengers are typically not engaged in driving tasks. As such, there is a need to pin down the moments for driving-related information during NDRTs.
As a result, the researchers are able to effectively identify the timing as well as the frequency of the passenger’s demands for explanations. The model also recognized the specific explanations that passengers want during driving situations.
The researchers then used their approach to develop a machine (ML) learning model that forecasts the best time to offer the passenger an explanation. They also performed a city-wide modeling to generate textual explanations based on different driving locations.
The preliminary analysis, according to the study, indicates the model’s potential for determining the timing of passenger’s demand for in-vehicle explanations. Meanwhile, the dataset can be used to generate textual explanation content relevant to environmental, driving-related, and passenger-specific contexts.
“Our research lays the groundwork for increased acceptance and adoption of autonomous vehicles, potentially reshaping urban transportation and personal mobility in the coming years.”
– Prof. Kim
Companies Advancing Automated Driving Solutions
Now, let’s take a look at companies that are shaping the future of automated vehicles and are also positioned to take advantage of the advancements in explainable AI.
In the AV sector, General Motors (GM +0.92%) has developed Cruise for driverless rides while Ford Motor is making its move in this direction via Escape Hybrid.
Then there’s NVIDIA (NVDA -1.06%), whose DRIVE platform offers a family of hardware and software tools for autonomous vehicle development. Amazon (AMZN +2.38%) is also interested in autonomous vehicle technology via Zoox, which has started testing its driverless cars ahead of its launch next year. The likes of Uber (UBER -0.4%) and Lyft (LYFT -3.33%), which have a ride-sharing network, can also benefit from advancements in fostering trust and safety in self-driving vehicle services.
Now, two prominent names in the autonomous vehicle market that you may find investment-worthy are:
1. Waymo (GOOGL -1.32%)
In the world of autonomous vehicle development, Waymo is making a lot of progress. This Alphabet subsidiary focuses on self-driving technology and passenger-centric features.
Late last month, the company announced closing a $5.6 billion oversubscribed investment round led by parent company Alphabet, with other participants including existing investors and private equity firms Fidelity, Tiger Global, Andreessen Horowitz, Perry Creek, Silver Lake, and T. Rowe Price.
The funds will be used to expand its ride-hailing service “Waymo One” to more US cities and improve AI-powered “Waymo Driver.” Most recently, the company launched its robotaxis in Los Angeles, which means anyone in the city can call a driverless cap through the Waymo One app. This robotaxi has already been rolling around Phoenix for four years now and has been in San Francisco since last year. Meanwhile, in Austin and Atlanta, Waymo has added its AVs to Uber’s platform, allowing customers to hail its vehicle from the Uber app.
Google actually began working on self-driving cars over a decade and a half ago when Waymo’s was just a secret project. The tech giant’s self-driving cars have reportedly logged more than 20 million miles without any major accidents.
Alphabet Inc. (GOOGL -1.32%)
The $2.2 trillion market cap giant’s shares are currently trading at $180.91, up 30% this year. It has an EPS (TTM) of 7.54, a P/E (TTM) of 24.09, and a dividend yield of 0.44%. For 3Q24, it reported net sales of $2.93 billion and $702 million in cash flows from operations.
For Q3 2024, Alphabet reported a revenue of $88.27 billion, an increase of 15% year-over-year. Its cloud revenue surged 35% from a year ago to record $11.35 billion this quarter, driven by AI offerings.
AI has been garnering a lot of attention among users and companies alike, with Google attracting new customers, getting larger deals, and seeing increased adoption thanks to AI. So, naturally, the company continues to “invest in state-of-the-art infrastructure” to support its AI efforts.
2. Tesla Inc. (TSLA +0.59%)
Founded by Elon Musk, Tesla is known for its electric vehicles, which offer Autopilot as a level 2 automation. Autopilot is a standard feature on every new Tesla, with each of its vehicles equipped with multiple cameras and vision processing for an extra layer of safety.
Then there’s Full Self-Driving (FSD), which adds semi-autonomous navigation. Both Autopilot and FSD are intended for use with a fully attentive driver.
While Autopilot includes functionality like traffic-aware cruise control and autosteer, FSD (supervised) offers additional features, including navigation on autopilot, autosteer on city streets, auto lane change, auto park, summon and smart summon, traffic control, and stop sign control.
Tesla vehicles also come with several active safety features that allow them to detect cars or obstacles, impending collision warnings, side collision warnings, and blind spot monitoring, among others, to assist drivers.
The automaker, however, is currently facing scrutiny from the National Highway Traffic Safety Administration (NHTSA), which expressed dissatisfaction with Tesla’s promotional language on social media regarding its FSD feature. The agency believes Tesla’s messaging could promote the unsafe usage of the system and has requested the company to reconsider its communication strategy about FSD capabilities.
This comes on the back of a tragic incident in which a woman was struck by a Tesla, which was operating in FSD mode, raising questions about the system’s ability to handle challenging environmental conditions.
NHTSA has given Tesla a Dec. 18 deadline to respond to its questions concerning FSD’s “potential failure to perform, including detecting and responding appropriately in specific situations where there is reduced roadway visibility that may limit FSD’s ability to safely operate.”
With a market cap of $1.05 trillion, Tesla shares are currently trading at around $340, up 32.2% year-to-date (YTD). It has an EPS (TTM) of 3.65, a P/E (TTM) of 90.07, and an ROE (TTM) of 20.65%. Its Debt To Equity (MRQ) is 11.01%.
Tesla, Inc. (TSLA +0.59%)
For 3Q24, the company reported $23.35 billion in revenue and a net income of $2.17 billion. The profit margins increased by $739 million in automotive regulatory credit revenue as a result of regulators requiring automakers to sell a certain number of low-emission vehicles or buy credits from the likes of Tesla, who exclusively builds such vehicles and, as a result, have those credits in excess.
In this quarter, the automaker manufactured 470,000 vehicles and delivered 463,000 vehicles. Recently, it also unveiled a robotaxi and robovan.
Click here to learn all about Tesla.
Conclusions
The vast expanding world of autonomous vehicles points to a future of increased mobility, reduced traffic congestion, more convenience, and improved safety.
While the global autonomous vehicle market size is projected to grow to $13,632.4 billion by the end of this decade, the market of autonomous driving software, which is integral to AVs, is expected to increase from $1.8 billion in 2024 to $7 billion by 2035.
The growing demand for efficient and safe transportation solutions is the reason behind the growth of the autonomous driving software market. As AVs gradually become popular and gain adoption, automotive manufacturers are required to incorporate safety technologies. Here, autopilot driving software ensures vehicle safety through algorithms and real-time data processing.
Now, to realize the future of fully autonomous vehicles, we are going to need more than just technological advancement. Earning passenger trust is critical to achieve widespread adoption. With solutions like TimelyTale, which focuses on timely and related explanations, passenger concerns can be addressed better, and trust fostered, hence creating a more human-centered approach to autonomous driving.
Innovations like these are important to move all that much closer to a future where self-driving vehicles are smoothly integrated into our everyday lives, in turn transforming urban mobility.
Click here to learn how autonomous taxis will generate up to $4 trillion by 2027.
