UNIT I AI for Driverless systems Lecture 8 Hrs

Look: The MEMEX Reloaded, Inside a Search Engine, Google and the Mind, Deeper and
Darker.
The Robotic chauffeur: Getting to driverless, A Cure for the deadliest disease, Seven delaying
Myths, The timeline.

1.1 The MEMEX Reloaded

In 1945, Vannevar Bush envisioned the MEMEX, a device designed to emulate and augment
human memory by storing all personal books, records, and communications for quick and
flexible access. This idea foreshadowed today's internet and web search capabilities. Bush's
MEMEX was modeled on human memory, emphasizing associative trails and dynamic access,
even though he acknowledged the challenge of matching the mind's speed and flexibility.

The modern web, particularly through Google, mirrors Bush's vision by offering vast, easily
accessible information, thus surpassing human memory in permanence and clarity. This vast
repository, known as "big data," powers various web-intelligence applications, enhancing our
cognitive abilities. For instance, tools like Google Glass enable users to instantly retrieve
information about objects and landmarks by simply looking at them. Despite potential privacy
concerns, such technology underscores the significant strides made towards creating a global,
collective memory accessible to all.

Search engines, handling billions of queries daily, have revolutionized information access and
cognition. They reduce the need to memorize facts, allowing instant retrieval of information.
However, studies suggest that the internet might be diminishing our deep reading and
comprehension skills, encouraging a skimming approach instead. Despite this, the ability to
connect disparate pieces of information is likely enhanced, mirroring how our brains correlate
memories.

Today's web search capabilities allow users to discover previously unknown connections
between people, ideas, and events, much like Bush's MEMEX envisioned. This augmented
recall aids various professions by providing quick access to relevant information, facilitating
better decision-making and problem-solving. In essence, web search augments our cognitive
abilities by serving as an external memory bank.

Google's founders, Larry Page and Sergey Brin, view search as a step towards artificial
intelligence, envisioning a future where search engines are as smart as, or smarter than,
humans. This vision aligns with enhancing our ability to connect the dots, potentially allowing
computers to autonomously derive connections from vast datasets, thereby advancing
automated reasoning.

In summary, the evolution from Bush's MEMEX concept to today's web search demonstrates
a significant enhancement in our ability to access and utilize information, fundamentally
augmenting human cognition and paving the way for more intelligent systems.

1.2 Inside a Search Engine

Arthur C. Clarke's adage, "Any sufficiently advanced technology is indistinguishable from

magic," aptly describes the sophisticated operations behind internet search engines, particularly
Google. This technology is powered by an extensive network of over a million servers,
dwarfing the largest banks' server networks, which max out at around 50,000 servers each.

The Core of Search Technology: Indexing

The fundamental concept of searching data has been central to computer science since its
inception. Similar to how a book index helps locate information efficiently, search engines use
indexing to store and retrieve data. Google's servers continuously crawl and index billions of
web pages, storing each word or phrase along with the URLs of the pages where they appear.
This massive index includes entries for every conceivable word and phrase, making it a
comprehensive resource for search queries.

Efficient Searching: Binary Search Algorithm

When a user types a search query into Google, the search engine employs a binary search
algorithm to quickly locate the query within the vast index. This algorithm operates by
repeatedly halving the search space, significantly reducing the number of steps needed to find
the desired word. For example, searching an index of a million entries requires at most 20 steps.
This efficiency enables Google's servers to handle billions of queries per second, providing
near-instant search results.

The Challenge of Sorting

Creating the sorted index is a complex task involving sorting a massive list of words. Unlike
searching, which can be done quickly, sorting requires more steps as it involves organizing the
entire dataset. Google's solution involves advanced algorithms and substantial computing
power to manage and update the index continuously.

The Role of PageRank

Google's breakthrough in search technology was the introduction of PageRank, an algorithm

that determines the importance of web pages based on the link structure of the web. PageRank
assesses a page's importance by considering how often it is likely to be visited by a random
surfer, influenced by the number and quality of incoming links. This global computation
ensures that search results are ranked by relevance and importance, greatly enhancing user
experience.

Managing Big Data

The vast size of the web index and the dynamic nature of web content necessitate continuous
updates. Google employs parallel computing innovations, such as the map-reduce paradigm,
to handle this enormous computational task. This approach not only ensures the efficiency of
the search engine but also influences data management technologies across various industries.

The Magic Behind Instant Search

Google's real-time search capability is a combination of efficient indexing, advanced

algorithms, and extensive computing resources. By storing and replicating the web index across
millions of servers, Google provides rapid search results, often displaying them even before
the user finishes typing the query. This seamless experience is the result of years of innovation
and optimization in handling big data and search technology.

In essence, the seemingly magical performance of search engines like Google is grounded in
fundamental computer science principles, advanced algorithms, and massive computational
infrastructure, making it a cornerstone of modern information retrieval.

1.3 Google and the Mind

Google’s sophisticated search technology relies heavily on its massive computing

infrastructure to maintain an accurate index of billions of web pages. The PageRank algorithm,
which determines the relative importance of each web page, plays a crucial role in delivering
high-quality search results. This process of ranking pages by importance parallels how humans
assign importance to memories, raising interesting questions about the relationship between
search technology and human cognition.

The PageRank Algorithm and Human Memory

Google’s PageRank operates by analyzing the link structure of web pages to determine their
importance, a process that resembles how humans might prioritize memories. A study at Brown
University explored this connection by testing whether the PageRank algorithm could predict
human memory recall. Researchers used a semantic model, where words are linked by
associations, and found that PageRank could successfully rank words in a way that matched
human recall patterns. This suggests that PageRank might offer insights into how human
memory works, although it’s not conclusive evidence of the brain using a similar mechanism.

Human Memory and Search Technology

Despite these insights, human memory and search engines differ significantly. Human memory
operates by associations, dynamically forming connections based on experiences, whereas
search engines rely on static hyperlinks. Moreover, human memory often recalls a single
relevant memory or a small set of related memories, unlike the lengthy lists generated by search
engines.

Limitations and Adaptations

Google’s search technology has to adapt continuously to the changing web, where new pages
may have fewer hyperlinks. This impacts the effectiveness of PageRank, necessitating the use
of additional data like browsing histories and user behavior to refine search results. The concept
of associative memories, which dynamically form links based on similarities, contrasts with
the static nature of web links, and models like Sparse Distributed Memory (SDM) explore these
differences further.

Challenges of Near-Duplicates

One significant challenge in search technology is handling near-duplicate documents, a

problem that grows exponentially with the number of documents. Techniques like Locality
Sensitive Hashing (LSH) help address this issue by identifying similar items without
exhaustive comparisons, making the process feasible even for vast datasets.
Conclusion

While Google’s PageRank and other search technologies provide valuable insights into aspects
of human cognition, the complexity of human memory remains far beyond the capabilities of
current search algorithms. The continuous evolution of search technology and the interplay
between web use and memory highlight the dynamic nature of both fields, with each
influencing the other in unexpected ways.

1.4 Deeper and Darker

Surface Web vs. Deep Web

Google's search capabilities give an impression of accessing all of the world's information, but
this is not the case. The indexed "surface web" is just a fraction of the "deep web," which
includes data hidden behind web services such as travel booking systems, private user data,
and dynamically generated content. The deep web is vast, containing potentially trillions of
pages, significantly larger than the surface web's 50 billion pages.

Efforts to Index the Deep Web

Search engines are attempting to index parts of the deep web. Google, for example, uses
automated techniques to interact with forms and queries to capture and index deep web content.
Other approaches, such as those by Kosmix, involve classifying and querying popular web
services on-the-fly. This area remains an active field of research and innovation.

Beyond the Web: Personal Data

Vast amounts of data are stored in databases not directly connected to the web, such as financial
records, telecommunication logs, legal documents, and travel records. The challenge lies in
accessing and correlating this data efficiently, a task made increasingly possible with
advancements in search technology and big data analytics.

Surveillance and Privacy Concerns

Post-9/11, there has been a rise in government surveillance and data correlation to prevent
terrorism. Programs like the US's Total Information Awareness (TIA), though controversial
and short-lived, aimed to integrate multiple databases for comprehensive monitoring. While
such initiatives have had success in preventing attacks, they raise significant privacy concerns
and the potential for misuse.

Counter-Terrorism and Real-Time Data

Real-time access to data, such as Call Data Records (CDRs) and triangulation information from
cell towers, could significantly enhance counter-terrorism efforts. The example of the 2008
Mumbai attacks illustrates how timely access to such data could have provided actionable
intelligence, potentially altering the course of events.
Na onal Intelligence Grid (NATGRID)

India's NATGRID aims to connect various public databases to support counter-terrorism. The
project envisions "Google-like" searches across multiple data sources, from structured data like
CDRs to unstructured public information. Despite technical feasibility, the initiative faces
challenges in inter-agency cooperation, privacy concerns, and legal issues.

Impact of Web Search on Expecta ons and Behavior

The ubiquity and efficiency of web search have shaped public expectations for instant access
to information, influencing areas like law enforcement and personal data management. The
dependence on search engines is changing how people think and process information,
potentially affecting memory and cognition.

Ethical and Future Considera ons

The advancement in search technologies, while providing numerous benefits, also poses ethical
dilemmas related to privacy and surveillance. As these technologies evolve, their impact on
society, personal freedoms, and the balance between security and privacy will continue to be
critical issues. The interaction between humans and technology, with systems increasingly
mirroring intelligent behavior, raises profound questions about the future of information access
and control.

The Robotic Chauffeur

Evolu on of Cars and Current Challenges

Cars have remained largely unchanged in their basic design for nearly a century, while
advancements in technology have revolutionized many other fields. The era of human-driven
cars is transitioning into one of autonomous vehicles, thanks to recent developments in
robotics and artificial intelligence (AI). This transformation promises safer, cleaner, and more
convenient transportation.

Benefits and Drawbacks of Human-Driven Cars

Human-driven cars have reshaped urban landscapes, offering freedom and mobility, but at
significant costs, including traffic accidents, urban congestion, and pollution. With roughly a
billion cars on the road, these issues are pervasive, yet cars remain integral to modern life.

Advantages of Autonomous Vehicles

Autonomous vehicles can mitigate many of these problems. They rely on intelligent software
to handle repetitive driving tasks and respond to clear hazards, which can reduce accidents
and improve traffic flow. Unlike more complex mobile robots, cars benefit from simpler
movement dynamics, making the development of driverless technology more feasible.
Technological Hurdles

The development of autonomous cars has been delayed by the need for reliable software
capable of handling rare, unpredictable events, known as "corner cases." Unlike factory
robots operating in controlled environments, driverless cars must navigate the chaotic and
unpredictable real world, which requires sophisticated perception and situational awareness.

Advances in Machine Vision and Percep on

Machine vision has historically been a major barrier due to its reliance on rigid, rule-based
systems that struggled with novel situations. Recent breakthroughs in deep learning have
revolutionized this field, enabling software to accurately classify objects and handle
unstructured environments. Deep learning allows autonomous systems to learn from vast
amounts of data, improving their ability to recognize and respond to diverse scenarios.

The Future of Autonomous Vehicles

Deep learning has catalyzed advancements in AI, attracting significant investment from
automotive and tech companies. As these technologies continue to evolve, they will not only
transform cars but also impact the broader field of robotics, leading to more agile and
perceptive machines. This technological evolution mirrors the Cambrian Explosion in
biological history, where the development of vision led to a rapid diversification of life
forms.

Implica ons and Predic ons

As robots gain perception capabilities, they will move beyond controlled industrial
environments into more diverse and dynamic roles. This "robotic Cambrian Explosion" will
see the emergence of a wide variety of robotic forms and functions, each adapted to specific
tasks and environments. The advent of autonomous vehicles is just the beginning of a broader
revolution in robotics and AI, promising significant changes in how we live and work.

Getting to Driverless

The Vision of Driverless Cars

In the future, streets and highways will be filled with driverless cars navigating seamlessly,
akin to a school of fish. These vehicles will move efficiently and safely, avoiding collisions
and optimizing traffic flow, all while providing various services, from passenger transport to
deliveries.

The Path to Autonomous Vehicles

The transition from human-driven to driverless cars will be both gradual and sudden.
Advances in technology, as predicted by Moore's Law, have made sensors, data processing,
and computing power more reliable and affordable. Driverless cars typically use digital
cameras, radar, lidar, GPS, and inertial measurement units to create a comprehensive digital
model of their surroundings.
Technological Readiness and Challenges

While the technology for autonomous vehicles is almost ready, societal and regulatory
challenges remain. Resistance from consumers is expected to be minimal, especially among
younger generations who are more open to driverless cars. Surveys indicate a strong
willingness to adopt autonomous vehicles if they prove safe and efficient.

Regulatory and Legal Hurdles

The primary barriers to widespread adoption are regulatory and legal. Governments need to
establish clear guidelines, update liability laws, and restructure insurance policies to
accommodate driverless cars. Proactive government involvement is crucial to address these
issues and ensure the safe integration of autonomous vehicles into society.

Safety and the Urgency of Adop on

The human cost of delaying autonomous vehicles is significant. Traffic accidents, primarily
caused by human error, result in numerous fatalities and injuries each year. Driverless cars
have the potential to reduce these incidents drastically. However, the transition requires
overcoming regulatory inertia and societal resistance.

Future Implica ons

As autonomous vehicle technology matures, it will revolutionize transportation, reducing

accidents, optimizing traffic, and lowering emissions. The shift towards driverless cars will
also reflect broader technological trends and societal changes, ultimately leading to safer and
more efficient transportation systems worldwide.

A Cure for the Deadliest Disease

The Deadly Toll of Automobiles

Automobiles have caused a staggering number of deaths since their invention. Each year,
around 1.2 million people die globally in car accidents, a death toll comparable to ten
Hiroshima-scale atomic bombs annually. This number is nearly as high as the combined
deaths from war, violence, and drugs.

The Lack of Response

Despite the high fatality rate from car accidents, society shows a grim acceptance of these
deaths. There are no significant governmental campaigns or public outcries comparable to
those against other causes of death, such as drugs or violence. After major accidents, there is
rarely sustained outrage against car manufacturers.

The Poten al Solu on

A solution to drastically reduce car accident fatalities exists: replacing human drivers with
intelligent software and sensors. Autonomous vehicles could be a major public safety
advancement, akin to a cultural "Apollo moment." If 90% of cars in the U.S. were
autonomous, driving-related deaths could drop from 32,400 to 11,300 annually, according to
a study by the Eno Center for Transportation.

New Risks and Overall Benefits

While driverless cars won't eliminate all risks—such as the potential for hacking or new
forms of human misbehavior—the benefits outweigh these concerns. Autonomous vehicles
will likely lead to safer roads, reduced traffic jams, and lower air pollution. Additionally, they
will offer more personal mobility options, especially for the elderly and impaired, who
currently face significant transportation challenges.

Enhancing Quality of Life

Driverless cars will provide significant quality-of-life improvements. They will ease the
burden of tedious commutes and grant more people access to safe, convenient transportation.
This can open up new opportunities for where people live, work, and engage in recreational
activities.

In summary, autonomous vehicles hold the promise of significantly reducing the death toll
from car accidents and improving overall transportation efficiency and accessibility.

Seven Delaying Myths

The adoption of driverless cars faces opposition rooted in seven pervasive myths, which
hinder their progress and the numerous benefits they promise.

1. Autonomous Driving Will Evolve from Driver-Assist Technology: Gradually

transitioning through driver-assist features is unsafe and impractical. Research shows
that humans become less attentive when machines share control, making emergency
situations more dangerous.
2. Technological Progress is Linear: The belief that driverless car technology will
develop at a consistent pace is incorrect. The technology will advance rapidly due to
Moore's Law and deep learning, allowing cars to share and learn from vast amounts of
driving data collectively.
3. The Public is Resistant: Contrary to the notion that people love driving, most find it
tedious. Surveys show a strong willingness to adopt driverless cars, especially if they
reduce travel time and enhance safety.
4. Driverless Cars Require Extensive Infrastructure Investment: This misconception
arises from outdated visions of connected cars. Driverless cars mainly need clear lane
markings and rely on advanced machine-vision technologies and digital maps, not
expensive infrastructure.
5. Driverless Cars Represent an Ethical Dilemma: Ethical concerns over decision-
making in accidents overlook the fact that humans already make such decisions.
Driverless cars will handle these situations similarly, but potentially with more
consistency and safety.
6. Driverless Cars Need a Nearly Perfect Driving Record: The adoption of driverless
cars should begin once they are safer than human drivers, not perfect. A safety record
twice as good as the average human driver is a reasonable threshold to start saving
lives.
 7. The Adoption of Driverless Cars Will be Abrupt: Adoption will be gradual. Initial
use will occur in controlled environments like resorts and campuses, with gradual
integration into public roads as human-driven cars phase out.

Addressing these myths is crucial for embracing driverless technology, which promises to
reduce fatalities, pollution, and enhance mobility.

The Timeline

The timeline for the adoption of driverless cars is complex and non-linear, varying by
environment and region. The transition will be gradual, influenced by different stages of
adoption and regulatory approvals.

Key Points:

1. Special Environments First:

o Autonomous vehicles are already in use in specialized se ngs like mines and farms.
o Ini al city adop on will involve low-speed shu les in controlled environments (e.g.,
airports, resorts).
o Over me, these shu les will operate at higher speeds and greater ranges,
eventually moving onto city highways.
2. Geographical Variations:
o Different countries and states will adopt driverless cars at varying paces.
o Legaliza on and regulatory acceptance will also differ, influencing the adop on
meline.
3. Gradual Transition:
o Achieving full autonomy (100% of cars on the road being autonomous) could take up
to a century.
o The high cost and slow turnover of cars mean the transi on will span several
decades.
4. Projections:
o Automo ve and transporta on experts predict significant adop on post-2025.
o By 2035, around 10% of new cars sold are expected to be autonomous.
o Near-complete adop on is an cipated by 2050, with most new vehicles being
autonomous.
5. Staged Approach:
o Car companies are promo ng a gradual introduc on of autonomous features.
o Fully autonomous cars are expected to follow, with tech companies like Google and
Tesla leading the charge.
o Elon Musk predicts true autonomous driving in 5-6 years from 2014, with regulatory
approval following 2-3 years later.
6. Regulatory and Adoption Challenges:
o Without a federal mandate, adop on may mirror the gradual shi to automa c
transmissions, taking decades.
o Even with early legaliza on, it could take 20 years to phase out all human-driven cars
due to their long lifespan.
7. Human-Robot Coexistence:
o Humans and autonomous vehicles will share roads for many decades.
o The transi on will involve significant changes in infrastructure, technology, and
societal habits.
Conclusion

The journey to driverless cars is a gradual, multifaceted process influenced by technology,

regulation, and societal acceptance. While the full transition will take decades, the benefits of
autonomous vehicles—safety, efficiency, and convenience—make the effort worthwhile.

A Driverless World

Imagine a future where all the world's cars are reliable, driverless vehicles. The first
noticeable change would be the silence, as there would be no need for honking or sirens with
no human drivers. The streets would be filled with small, golf cart-like vehicles, some
carrying passengers and others completely empty. Occasionally, larger RV-sized vehicles,
perhaps functioning as mobile offices, would pass by. Human-driven cars would be rare, and
autonomous vehicles would give them a wide berth to ensure safety.

To summon a taxi, you would simply press a button on your phone. Within minutes, a
driverless taxi would arrive. If you opted for a ride-share, the pod might already have
passengers headed in the same direction, reducing your fare. The interior of these pods would
be basic and utilitarian, designed for easy cleaning with minimal features. They might
resemble a well-worn subway car, with litter and graffiti, and the taxi's security camera often
rendered ineffective by simple obstructions.

In this new era, traditional anonymous taxi rides paid in cash would be obsolete. Your pod
would recognize you upon entry, using your online browsing and shopping histories to
suggest routes that pass by your favorite stores. The ride itself would be relaxing and
impersonal, free from the pressure to engage in small talk with a driver. You would trust the
pod's software to take the most efficient route, eliminating the need to check the route on
your phone.

Upon arrival at your destination, the fare would be automatically deducted from your
account, with no need to tip. Sharing the pod and enduring advertisements would make the
ride affordable. Although the ride might not be particularly clean or comfortable, the ease and
convenience would be akin to taking an elevator, highlighting the utilitarian and efficient
nature of this new mode of transportation.

Friction-free Personal Mobility

The potential impact of driverless cars on traffic congestion and the environment is a topic of
debate. In an optimistic scenario, autonomous vehicles could enhance the efficiency of urban
transportation systems, reduce private vehicle ownership, and consequently lower congestion
and carbon emissions. Conversely, the convenience offered by driverless cars might lead to
increased vehicle miles traveled annually, resulting in a larger carbon footprint.

Economists refer to the unforeseen reduction of expected gains from new technologies due to
increased usage as the rebound effect. It is uncertain whether driverless cars will increase
traffic and miles driven each year. Research suggests a mixed outcome. For instance, a
simulation by Luis Martinez of the International Transport Forum predicts that shared
autonomous taxis could replace public transportation and reduce the number of vehicles on
the streets by 90%, although the miles traveled per vehicle might increase slightly.
The University of Michigan Transportation Research Institute supports these findings,
suggesting that driverless cars could reduce household car ownership from two to one
vehicle. However, this could result in the remaining car being used 75% more frequently.
While this increased usage might negate some benefits, overall mileage per household could
still decrease, potentially lowering environmental impact.

The environmental impact of driverless cars could be significant if convenience leads to more
miles driven. Autonomous cars might need to shuttle back and forth between destinations and
safe parking spots, increasing overall mileage and contributing to traffic congestion and
pollution. This is particularly concerning for densely populated cities in developing countries,
where air quality is already a major issue.

The lifespan of driverless cars could also affect their environmental impact. If driverless cars,
like New York taxis, accumulate high mileage quickly, they might need to be replaced more
frequently, leading to more waste. However, advancements in technology could mitigate
these effects. For instance, smoother driving by autonomous vehicles could reduce wear and
tear, potentially extending vehicle lifespan and decreasing emissions.

Technological improvements could alleviate some negative effects of driverless cars. For
instance, platooning, where vehicles drive closely together to reduce wind resistance, could
save fuel and reduce congestion. Additionally, lightweight car bodies and efficient driving
patterns could further enhance fuel efficiency. Driverless cars are also likely to use electric
engines, which could reduce their environmental impact if coupled with widespread charging
infrastructure.

Driverless cars could also improve urban living by reducing the need for parking spaces and
eliminating the cruising that drivers do in search of parking. This could lead to more efficient
use of urban space and reduce traffic congestion. In conclusion, while driverless cars offer
many potential benefits, their overall impact will depend on how they are integrated into
society and managed to balance convenience with environmental sustainability.

Parking

The charm of a city often lies in its pedestrian culture, with bustling streets filled with people
contributing to a vibrant atmosphere and economic activity. In contrast, cities dominated by
parking lots lack this charm, as these areas were designed with cars in mind rather than
pedestrians. This walkability factor is absent in many newer cities that have developed
around the needs of automobiles. Parking lots do not enhance a city's walkability or appeal,
and instead contribute to traffic congestion and pollution as people circle in search of
parking.

Parking expert Donald Shoup highlights the significant hidden costs of searching for on-street
parking, which include increased traffic congestion, accidents, fuel waste, air pollution, and
degradation of the pedestrian environment. On average, cars are parked 95% of the time,
consuming valuable urban space. A typical parked car occupies about 14 square meters, and
even more when including access roads. This results in a substantial amount of urban real
estate being dedicated to parking.

Most cities have zoning ordinances that mandate the provision of parking spaces for new
developments, contributing to the prevalence of parking lots. These ordinances were
originally intended to alleviate the frustration of finding parking, but they have inadvertently
led to modern cities being riddled with spaces devoted to storing cars. Parking coverage rates
in cities reveal the extent of this issue, with some cities dedicating significant portions of their
downtown areas to parking.

If driverless cars eliminate the need for downtown parking, cities could repurpose these areas
for more human-friendly uses. However, this transition poses challenges. City planners must
carefully consider how to transform former parking spaces into valuable urban assets. Factors
such as the local economy, population density, and municipal budget impacts from lost
parking revenue need to be addressed.

An example of the unintended consequences of efficient parking can be seen in Los Angeles.
The city built an expensive underground parking garage for Disney Hall, expecting parking
revenues to cover the costs. However, the convenience of parking directly under the hall
meant that visitors did not spend time or money in the surrounding area, detracting from the
vibrancy of downtown. In contrast, San Francisco's Davies Symphony Hall benefits from
having a smaller, slightly inconvenient parking lot, encouraging visitors to walk and engage
with the local area.

Driverless cars will reduce the need for urban parking lots, which could lead to more vibrant
downtown areas. However, the convenience of being dropped off directly at destinations
might also reduce pedestrian traffic, impacting local businesses. As driverless cars render
parking lots obsolete, cities will need to navigate the complex process of repurposing these
spaces while maintaining vibrant urban centers.

Commuting

The greater metropolitan areas of cities such as Mumbai, Mexico City, and Shanghai are each
home to more than 20 million people. By 2050, the global urban population is projected to
nearly double from today's 3.3 billion to 6.4 billion. These megacities will expand
extensively, filling their streets with cars as citizens struggle with daily commuting
challenges.

As urban populations grow, efficient space utilization becomes critical. With the advent of
driverless cars, city planners have the opportunity to repurpose urban parking lots into parks
and affordable housing. Another improvement could be making commuting more
manageable.

Currently, people spend significant time commuting. In the United States, the average
commute is about 30 minutes each way, totaling an hour daily, typically driven alone. New
forms of transportation can alter perceptions of distance. In the 1950s, cars enabled
convenient commutes into cities, prompting suburban living. My grandparents, like many
New Yorkers, moved from Manhattan to Queens, motivated by various factors, including
affordability, commute time, and school quality.

Driverless cars will simplify commuting, offering more housing options. With less congested
streets and increased residential spaces, more people might move to revitalized downtown
areas. Those preferring the countryside could enjoy a manageable commute. Traditional
commuting centers might become less desirable.
Driverless cars will also reduce transportation costs. Research from Columbia University's
Earth Science Institute modeled transportation costs with fleets of driverless shared vehicles
in Manhattan, Ann Arbor, and a small town in Florida. The study concluded that on-demand
autonomous vehicles could significantly lower transportation costs. Ann Arbor residents
could see a 75% cost reduction per mile, New Yorkers could pay 50 cents per mile instead of
$4, and small-town inhabitants could enjoy rides at about 46 cents per mile.

Take the Pod—Meet People

While conveniences are hard to resist, life's inefficiencies can sometimes foster positive
social interactions. One potential downside of cheap and efficient personal mobility could be
increased loneliness. According to the U.S. Census Bureau, one in four Americans lives
alone, and the number of people who report feeling lonely increases each year. Several
theories explain this rise in loneliness, including changes in family structures, work demands,
the isolating effects of television and the internet, and the superficial connections formed on
social media.

One reason people feel lonely is the lack of a "third space," a place that is neither work nor
home where people can socialize. In the past, professors enjoyed faculty clubs or tearooms,
and people went to church, bowling alleys, or local pubs. Modern life, however, offers few
options for such spontaneous social interactions. Many middle-class North Americans follow
a pattern of moving to suburban homes with private yards, resulting in long commutes and
busy weekends focused on home improvement and family time, leaving little room for social
activities.

Three factors are critical for making friends: physical proximity, repeated unplanned social
interactions, and an atmosphere where people can let their guard down. Work provides
physical proximity and repeated interactions but not the right environment for letting down
one's guard. Many people, therefore, have friends from their college years but not many from
later in life. College dorms provide the right mix of proximity, exposure, and a relaxed
atmosphere for forming friendships.

Driverless cars, like other technologies of convenience, will improve quality of life by easing
commuting and expanding living options. However, this convenience might eliminate another
traditional third space: public transit. Future city dwellers might miss the human contact of
subways, buses, and trains. An ad promoting public transportation once cleverly highlighted
that taking the bus helps people meet new friends, acknowledging that public transit might
not be faster or easier than driving but offers social benefits.

Time spent in the car has hidden value. For parents and children, car rides serve as a mobile
third space. A study found that parents spend an average of six hours and forty-three minutes
per week chauffeuring their kids to school and activities. This results in nearly thirty hours
per month of enforced togetherness, providing opportunities for bonding.

Parents often see chauffeuring as drudgery and might welcome driverless cars. However, this
convenience would eliminate valuable time spent interacting with their children. For
example, early morning drives to crew practice with my teenage son allowed us to bond over
music and conversations. A driverless car would have provided convenience but at the cost of
losing these precious moments.
Friction-free physical mobility could increase loneliness, but it might also make social
interactions easier. If going out is as simple as pressing a button on a phone, in-person
mingling could become as easy as online interactions. Driverless cars could be programmed
to match passengers based on age or interests, creating opportunities for new friendships.
Advertisements might someday promote driverless pods with the slogan, "Take a pod. Make
a friend."

Anatomy of Driverless Cars

1. High-Definition Digital Maps High-definition digital maps are essential for autonomous
vehicles to navigate safely and efficiently. These maps provide detailed information about
road geometry, traffic signs, lane markings, and other relevant features. Unlike standard
maps, HD maps are precise to within centimeters, allowing driverless cars to understand their
exact position and make informed decisions.

Key Features:

 Accuracy: Provides centimeter-level precision.

 Details: Includes road boundaries, intersections, traffic signals, and 3D
representations of the environment.
 Dynamic Updates: Can be updated in real-time to reflect changes in road conditions
and traffic.

2. Digital Cameras Digital cameras are used in driverless cars to capture images of the
surrounding environment. These images are processed to detect and classify objects such as
vehicles, pedestrians, traffic lights, and road signs.

Key Features:

 Object Detection: Identifies and tracks objects in the vehicle's vicinity.

 Color Recognition: Helps in recognizing traffic signals and road signs.
 Lane Detection: Assists in identifying lane markings and maintaining lane position.

3. Light Detection and Ranging (LiDAR) LiDAR systems use laser beams to create a high-
resolution 3D map of the environment. By measuring the time it takes for the laser beams to
return after hitting an object, LiDAR can determine the distance and shape of surrounding
objects with high accuracy.

Key Features:

 Precision: Provides detailed 3D mapping with high accuracy.

 Range: Can detect objects over long distances.
 Environmental Adaptability: Functions effectively in various lighting conditions.

4. Radio Detection and Ranging (Radar) Radar systems in driverless cars use radio waves
to detect the speed, distance, and movement of objects. Radars are particularly useful in
adverse weather conditions, such as fog, rain, and snow, where cameras and LiDAR might be
less effective.

Key Features:
  Speed Measurement: Accurately detects the speed of moving objects.
 All-Weather Capability: Performs well in poor visibility conditions.
 Long Range Detection: Can identify objects at considerable distances.

5. Ultrasonic Sensors (Sonars) Ultrasonic sensors use sound waves to detect objects at short
distances. These sensors are typically used for parking assistance and low-speed maneuvers.

Key Features:

 Proximity Detection: Ideal for close-range object detection.

 Parking Assistance: Helps in detecting obstacles during parking.
 Cost-Effective: Relatively inexpensive compared to other sensors.

6. Global Positioning Systems (GPS) GPS provides the geographical location of the vehicle
by receiving signals from satellites. This information is critical for navigation and helps the
vehicle understand its position on the map.

Key Features:

 Global Coverage: Provides location data anywhere on Earth.

 Navigation: Assists in route planning and real-time navigation.
 Integration: Works in conjunction with HD maps and other sensors.

7. The Inner Ear (IMU) The Inertial Measurement Unit (IMU), often referred to as the
"inner ear" of the vehicle, consists of accelerometers and gyroscopes that measure the car's
movement and orientation. The IMU provides data on acceleration, rotation, and tilt, which
helps in stabilizing the vehicle and improving navigation accuracy.

Key Features:

 Motion Detection: Measures acceleration and rotational rates.

 Stabilization: Enhances vehicle stability and control.
 Navigation Support: Improves accuracy of position and orientation data.

8. Drive by Wire Drive-by-wire technology replaces traditional mechanical and hydraulic

control systems with electronic controls. This technology is used to manage steering,
acceleration, and braking through electronic signals, enhancing precision and response times.

Key Features:

 Precision Control: Offers more accurate control of the vehicle's functions.

 Redundancy: Improves safety through multiple electronic systems.
 Integration: Easily integrates with autonomous driving systems.

These components work together to enable driverless cars to perceive their environment,
make decisions, and navigate safely. Each technology plays a crucial role in ensuring the
reliability and efficiency of autonomous vehicles.