City Life Turned on its Head
I wonder how many hours each of us has spent behind the wheel of a car. Really, how many days have we spent navigating to the store, school, or wherever place where the only practical mode of transportation was a car? A study published by Harvard Health found that the average American spends 101 minutes driving per day [1]. This metric is equivalent to about 3.5 years over an average lifespan–finding some way to return that time would be invaluable. Well, say hello to autonomous vehicles!
Autonomous vehicle technology has seen massive improvements in the last few years; many companies provide fully autonomous taxis on the road today. Even more impressively, these taxis are primarily deployed in major cities, such as LA and Berlin. Autonomous vehicles can genuinely change urban city life via the reutilization of time, space, and resources. However, much work must be done to facilitate trust between human society and fully autonomous vehicles.
High Level: How do Autonomous Vehicles Operate?
My first experience in an autonomous vehicle is imprinted in my mind. Anxiously sitting in the passenger seat of a classmate’s brand-new car, we were driving on the narrow and curved roads of Capitola, California, a town near Santa Cruz. I watched him take his hands off the wheel and let the bright red Tesla navigate the streets independently. I was wary, exerting all of my attention watching the wheel to ensure that it would keep us in the lanes–keep us alive. I don’t doubt that other people may have similar feelings when first stepping into a new technology such as this, especially if no one is behind the wheel to take over if things go awry. Now, I realize I was merely experiencing a Level 3 Autonomous vehicle.
Six different levels encapsulate the automation that a car can exhibit. At Level 0, a human driver is aided by safety systems in the vehicle, such as an alarming noise if the car goes out of its designated lanes; however, the driver has total control. On the opposite side of the spectrum, Level 5 is when the vehicle has complete control over ODD, the operational design domain. ODD refers to how an autonomous vehicle may change driving style given certain geographical, environmental, or lighting conditions [2]. See the figure below to gain insight into the features each level of automation includes.
Figure 1. Six Levels of Car Automation. Each level of car automation includes additional autonomous features [3].
My experience in the Tesla was Level 3: in this case, an automated driving system (ADS) performs either longitudinal or lateral control of the vehicle in question, as pictured above. ADS and ODD work together to inform any robotaxi how to navigate its current environment, such as staying within the lanes or stopping if a pedestrian enters the street. Complete automation, Level 5, obviously becomes the goal state for all AV providers. To get there, they must utilize various technologies to ensure the safety of the vehicle’s operations.
Inside an Autonomous Vehicle
As one can imagine, a wide variety of features must be developed for a vehicle to be fully autonomous. Some of these include automotive sensors and localization algorithms. Automotive sensors allow the car to self-sense, localize, and surround-sense; in simple terms, they allow the vehicle to understand its position relative to its current environment. Proprioceptive sensors are commonly used for self-sensing, as they allow for measuring the vehicle’s velocity, acceleration, steering angle, etc.; some example sensors for this use case would be odometers to measure traveled distance or gyroscopes for orientation and angular velocity.
Sensors used on the vehicle can generally be categorized as active or passive. Active sensors function by radiating electromagnetic waves and measuring the time before those waves return, allowing the car to understand its distance from relative objects. Sonar, radar, and “light detection and range” sensors are all deemed to be active. On the other hand, passive sensors allow the vehicle to understand its environment by measuring electromagnetic waves already present via light-based or infrared cameras.
An important distinction between the sensor types is their scalability. In highly congested traffic environments, too many active sensors nearby may cause AVs to sense each other’s emitted waves rather than their own, resulting in confusion and incorrect distance estimation. For this reason, more resources are currently being invested in improving vision-based (passive) sensors for autonomous environment orientation [4].
Another exciting aspect of autonomous driving is the requirement for real-time mapping to ensure successful navigation. Beginning with indoor mapping for mobile robots, localization and mapping algorithms have dramatically evolved to allow AVs to navigate the roads independently. Simultaneous Localization and Mapping (SLAM), the process of determining self-position while creating an accurate map, is computationally intensive; the vehicle must take in information and almost instantaneously understand how to navigate it. Engineers and researchers are developing algorithms to accomplish this task. At Carnegie Mellon University, a world-renowned university for innovation in technology and Computer Science, engineers and researchers are designing a technique to flag environmental obstacles based on their predicted motion and subsequently create a dynamic map (including these objects) to allow a vehicle to develop an accurate navigation plan [4].
With engineers and researchers devoting precious resources to innovating this technology, it is pertinent that society makes an effort to understand said technology. Without doing so, it will become increasingly difficult for autonomous vehicles to become prominent, especially in major cities where their incorporation could be immensely beneficial.
Bye, Congestion. Hello, Public Spaces.
Autonomous vehicles have the potential to drastically improve urban infrastructure, a challenge that the National Academy of Engineering has identified as pertinent for society to solve. The quality of urban form, which is the built and natural environment which includes infrastructure, prominently contributes to the livability of urban environments. Autonomous vehicles present an opportunity to drastically decrease the infrastructure required for transport. Specifically, this can be achieved via autonomous vehicles leveraging cooperative driving; here, each autonomous vehicle can communicate with other autonomous vehicles on the road, sending information including driving intention and current driving state. A study conducted at Fujitsu Laboratories of America—Fujitsu, a multinational information and technology company from Japan—designed and tested a strategy called Altruistic Cooperative Driving (ACD), where autonomous vehicles can identify congestion conditions amongst themselves and coordinate to solve them, reducing congestion on the road. Results after testing ACD show statistically significant improved speed and resolution of traffic congestion; these results demonstrate that although the testing was done on the ACD strategy specifically, any number of cooperative driving strategies can be potentially used to optimize driving conditions and traffic flow [5].
Taking this into account, widespread incorporation of AVs would also allow for redesigning cities to be not highway-centered but town- and people-centered instead. Highways and roads can be narrowed and shortened as cooperative driving technology optimizes traffic flow, giving precious space to people instead of cars. The image below depicts a repurposed public space.
Figure 2. Repurposing of Roads to Public Space. A multitude of people gather and connect in a public space, which used to be merely a means of transportation by car [6].
Converting even a few major roads in towns into public spaces can facilitate relationships within and strengthen the community, and bring greenery back into urban spaces.
Additionally, AVs give people their commute time back in a highly convenient way, in an ideal world, getting picked up and dropped off at the exact location they require. While public transport can provide similar benefits, one may argue that the average commuter could more confidently leverage the convenience and safety of an individual AV ride. With this, people could confidently start work before arriving at the office or finish another chapter in a book they never used to make time to read. As commute time becomes less important in the eyes of the individual, people may begin to find homes further away from cities, again reducing congestion and over-population in dense areas. However, it should be noted that this can also result in urban sprawl—the spread of urban development—which may lead to the consumption of more land and natural resources [7].
Consumer Wariness
Given this background, a reader of this article may think:
Well, this is awesome. Why aren’t we all using autonomous vehicles?
As I expressed earlier, it is pervasive for people to be wary and have little trust that a robotaxi or autonomous vehicle will safely get them to their destination. As a well-known provider of autonomous cars, Waymo, has been an independent autonomous vehicle provider since 2016. In 2020, it opened up a fully autonomous rider-only service to the entire Metro Phoenix area, and today, the company has expanded to San Francisco, LA, and Scottsdale. However, even with years of experience and little to no accidents, if there are any minor hiccups in the vehicle’s performance, society is quick to turn against the technology.
Around two months ago, on February 10th, 2024, a crowd in San Francisco surrounded a Waymo vehicle, graffitied it, and set it on fire. The image below depicts the state of the vehicle after the incident.
Figure 3. Burnt Waymo Vehicle in San Francisco. The vehicle was completely destroyed, with the entire roof melted off [8].
The final state of the vehicle, as pictured above, indicates the severity of the fire as well as the intensity of the event. Reporters speculate that this act of aggression came from prior malfunctions of autonomous vehicles. Namely, Waymo recently recalled software that resulted in two minor incidents in Phoenix [9]; more alerting, however, another autonomous vehicle provider, Cruise, was responsible for hitting and dragging a pedestrian 20 feet due to technical failures [10]. It should be noted that the accident began because another human-driven vehicle hit the woman on the other side of the street, propelling her in front of the Cruise vehicle. Notably, 94% of autonomous vehicle accidents are caused by human-related errors [4]. This raises the question of what the future of driving may look like—will there be any human-operated vehicles on the road?
It’s difficult to speculate, but addressing issues such as these and building a sense of trust between robotaxis and the general public will be an intensive journey, a prerequisite being that autonomous vehicle providers can craft technology that can navigate the roads better than a human can. Even though human drivers make statistically more mistakes than autonomous vehicles, there seems to be a prevalent idea that because these cars are autonomous, they must be perfect on the road. With AVs, however, the future of urbanization and city life seems to be drastically different (and improved) compared to current circumstances.
References
[1] H. H. Publishing, “Moderate exercise: No pain, big gains,” Harvard Health. https://www.health.harvard.edu/newsletter_article/Moderate_exercise_No_pain_big_gains
[2] Parekh, D.; Poddar, N.; Rajpurkar, A.; Chahal, M.; Kumar, N.; Joshi, G.P.; Cho, W. A Review on Autonomous Vehicles: Progress, Methods and Challenges. Electronics 2022, 11, 2162. https://doi.org/ 10.3390/electronics11142162
[3] Siddahantjain, “Six Levels of Vehicle Automation: Fully Manual to Fully Autonomous,” Medium, Aug. 03, 2023. https://medium.com/@siddahantjain50/six-levels-of-automation-fully-manual-to-fully-autonomous-2d528ea50c63
[4] J. Van Brummelen, M. O’Brien, D. Gruyer, and H. Najjaran, “Autonomous vehicle perception: The technology of today and tomorrow,” Transportation research. Part C, Emerging technologies, vol. 89, pp. 384–406, 2018, doi: 10.1016/j.trc.2018.02.012.
[5] N. Wang, X. Wang, P. Palacharla and T. Ikeuchi, “Cooperative autonomous driving for traffic congestion avoidance through vehicle-to-vehicle communications,” 2017 IEEE Vehicular Networking Conference (VNC), Turin, Italy, 2017, pp. 327-330, doi: 10.1109/VNC.2017.8275620.
[6] “Reclaim Street Space in the Driverless City,” Urban Design Forum, Feb. 26, 2018. https://urbandesignforum.org/reclaim-street-space-in-the-driverless-city/
[7] Fábio Duarte & Carlo Ratti (2018) The Impact of Autonomous Vehicles on Cities: A Review, Journal of Urban Technology, 25:4, 3-18, DOI: 10.1080/10630732.2018.1493883
[8] J. S. / Gizmodo, “A crowd set a Waymo self-driving car on fire in San Francisco,” Quartz, Feb. 12, 2024. https://qz.com/waymo-self-driving-car-fire-san-francisco-1851247761 (accessed Feb. 15, 2024).
[9] “Voluntary recall of our previous software,” Waymo. https://waymo.com/blog/2024/02/voluntary-recall-of-our-previous-software/ (accessed Feb. 15, 2024).
[10] How GM’s cruise robotaxi tech failures led it to drag …, https://www.reuters.com/business/autos-transportation/how-gms-cruise-robotaxi-tech-failures-led-it-drag-pedestrian-20-feet-2024-01-26 (accessed Feb. 15, 2024).
