The pace of technological breakthroughs in the last decade has revolutionized the way we live and lead our lives, and mobility is no exception. Mobility as a concept is no longer restricted to “trains, planes, and automobiles”. With advanced technologies, growing demand and cleaner energy considerations, the field has spawned a number of technology-driven disrupters.
The emergence of autonomous vehicles, passenger drones and other innovative driverless mobility solutions (collectively “AV(s)”) has transformed the traditional outlook and outreach of mobility. The key here is ‘automation’, which ranges from SAE[1] level 0 with no automation to SAE level 5 requiring no human intervention. These new tech-heavy, widespread and powerful AVs come with a myriad of benefits including in theory, higher traffic efficiency, 360 degrees vision, cleaner energy utilization, safer travel modes, increased access for people with disability and reduced mobility, amongst others.
In these concerning times of rising global temperatures, coupled with increasingly bad traffic around the world, the emergence of automated mobility solutions assumes greater significance. Smart automated mobility can solve many of the problems currently associated with mass transportation. This holds especially true for the South Asian region, which is bearing the brunt of unprecedented heatwaves[2], and also hosts five[3] out of the ten most densely populated countries, with people living in ‘very concerning’ levels of Air Quality Index[4].
However, despite the undoubted benefits, these innovative mobility means have not yet been tapped and utilised to the fullest where they are most needed. For instance in India, we do not have a driverless taxi drop us to our office, nor do we have our Amazon packages delivered via passenger drones. This holds true for most developing nations across the globe. The stark reality is that barring a few states in the U.S. and China, the rest of the world remains conspicuously underserved. Let alone truly advanced technologies such as passenger drones, that are currently in testing stages[5] or utilized for limited emergency or delivery services[6] in a few selected locales.
This calls for the need for a large-scale rollout of smart mass mobility solutions, which solves a lot of the prevalent burning issues. However, there are no roses without thorns! Other than the obvious cost, technology, access, and infrastructure issues (admittedly not insurmountable), AVs grapple with complex questions of law, which are summarised below.
Square pegs in a round hole?
Our legal systems governing transport are ‘geared’ to vehicles being driven by a human agent. And this is as much a human cognitive barrier as a legal one. As a mother to a toddler, the first question that crops up is can I let my child go to his school in an AV, with no driver?!
Existing legislations are not fit for the purpose, as most of the regulatory frameworks of motor or aerial vehicles worldwide are “driver-centric”, requiring the vehicle to remain in ‘effective control’ of the driver at all times. In fact, even the multi-country Vienna Convention on Road Traffic, 1968 (ratified mostly by advanced European countries); until as late as its amendment in November 2021, required all vehicles to have a driver[7].
In India, the extant legislation governing motor vehicles, The Motor Vehicles Act, 1988 inter alia mandates that motor vehicles are constructed and maintained in a manner that ensures that the vehicle remains under the effective control of the driver. In addition, it requires that the driver of the vehicle should not allow any person to stand, sit or place anything in a manner or position that hampers his/her control of the vehicle. Additionally, the Motor Vehicle Driving Regulations, 2017 (“Driving Regulations”) in India governs the conduct of the driver and inter alia specifies that the driver must ensure that any equipment in the vehicle does not obstruct their view or impair hearing capabilities and goes one step further to expressly mandate that the driver must not watch digital motion pictures, videos or use a hand-held device or any other communication device while driving (except for route navigation).
As noted, this is at its root a problem of conceptualisation under current laws; the absence of a driver is simply not contemplated in the current regulations. This complicates the case of AVs, as most AVs use various applications and artificial intelligence (AI), which may require active (dis)engagement of the driver. Such systems and applications may be viewed as a feature, which may (potentially) disrupt the driver's control and divert attention away from the road. Further, liability for breach of Driving Regulations is imposed on car manufacturers, as well as drivers. A car manufacturer or any other 'person' who causes or allows a vehicle to be driven in non-compliance with road safety standards may be subject to punishment.
In view of the above, it becomes imperative for the extant legislation, to be amended to tailor the intricacies, recognize, and specifically provide for driverless vehicles, with safety and other factors taken into consideration. In the author’s view, this calls for the enactment of specialised new legislation as one cannot redraft or reconcile a 1988 driver-centric laws to AVs easily, or at all. Motor regulations governing AVs will need to be drafted as a specific subset of traditional motor vehicle regulations.
Caveat Pedestrian: The Liability Conundrum
Last year, the United States Department of Transportation reported[8] that it received incident reports of 392 crashes by SAE Level 2 ADAS[9] vehicles and 130 crashes by SAE Level 3 to Level 5 ADS (autonomous driving system) vehicles between the period from June 2021 to May 2022. All these crashes, leading to injuries, fatalities and ultimately to the filing of lawsuits give rise to the next legal issue, that is, who shall be held liable in the event of an accident by an AV? Should it be the vehicle manufacturer, operator, software developer, passenger or insurer?
Let’s take, for example, the first AV fatality case on March 18, 2018, in Arizona, U.S. of one pedestrian, Ms Herzberg, who was crossing a four-lane road with her bicycle. The pedestrian was hit by Uber’s automated test vehicle, Volvo XC 90, which had Ms Vasquez, as the vehicle operator, occupying the driver’s seat with controls operated by the ADS of Uber Technologies, Inc. The crash led to the unfortunate death of the pedestrian.
The ADS data as per the report[10] by the U.S. National Transportation Safety Board, provided some insight into the accident – (a) ADS did not accurately classify Ms Herzberg as a pedestrian – it initially classified the pedestrian as a vehicle, then as an unknown object, and then as a bicycle, (b) the pedestrian was detected by the operator only 5.6 seconds before the collision, (c) the operator started steering left 0.02 seconds before striking the pedestrian, at a speed of 39 m.p.h., (d) as soon as the system detected that the crash was imminent, the design of the ADS prevented the emergency braking for crash mitigation alone and relied upon the operator to take control of the vehicle.
(Five years later, the vehicle operator continues to be tried on negligent homicide charges[11], for being distracted while operating the vehicle; criminal charges were not pressed on Uber due to a settlement with the deceased’s family.)
This case reflects the conundrum in the determination of liability – whether the automated system or the vehicle operator failed to prevent the crash? Add to it, the complexities arising on account of the usage of applications and AI in these AVs. Such applications and AI are merely carrying out the assigned functions and programs. Hence, can the software developer be held liable in case of an accident due to faulty AI?
The allocation and fixation of liability in accidents involving AVs are both highly complex and central to making AVs mainstream. Additionally, legislation should also provide for computation and quantum of compensation to the victims, commensurate to the degree of injury in case of accidents.
On the subject of liability, the Automated and Electric Vehicles Act, 2018 in the United Kingdom, provides that in the event of an accident by automated vehicles, the insurer will be held liable for damages, in case the automated vehicle is insured and if it is not insured, the vehicle owner will be held liable. Contributory negligence of the injured party also applies.
Big Motor is watching you!
AVs generate a significant amount of data, ranging from vehicle data to personal data to geolocation data. In the AV ecosystem comprising of vehicle manufacturers, operators, software developers, insurers, passengers and regulators, each will have different usage and obligations with respect to the treatment or utilization of the generated data. Such data may be commercially lucrative as not only it may give insights into the vehicle data, including location, routes, condition, functions, manner of driving and operation, but also provide personally identifiable data, including personal data, passenger preferences, etc. The collection, utilization, analysis, protection and storage of such data, especially personal data, is tasked with greater responsibility, due to privacy and security considerations. In India, currently, there are no specific laws regulating the collection of vehicular location data.
The basis of processing personal data is obtainment of consent in laws across the world. The question of express consent assumes importance, as interactive interfaces in an AV may not be an effective tool to communicate and obtain express consent from the consent giver. For instance, the digital interface may provide information about the vehicle, its safety, responsibility and liability on account of an accident, for which consent will have to be obtained. For such consent to have legal significance, the consent giver must fully understand these information before agreeing to be bound.
One also cannot ignore the lurking threat of cyberattacks. With the Internet of Things in play, a large number of devices are connected to one another, and a cyberattack may spiral into a catastrophe whereby hackers may end up having access to all data in the connected devices, including private networks.
Thus, the protection of individual’s privacy rights coupled with, cybersecurity and the establishment of a robust data protection legislation regulating the collection, usage, analysis, sharing, protection and storage of data with the following common data privacy principles in mind, becomes imperative for legislation on AVs to be legally sound and effective:
Express prior consent to be obtained from the consent giver at the time of data collection, with the option given to review, correct and/or refuse to provide such consent and in case, already provided, the option to withdraw previous consent;
Collection, usage, analysis, protection, and storage of personal data must be for lawful purposes connected with the data collector’s function or activity;
Privacy policy of relevant stakeholders must be in place and published;
No third-party transfer or disclosure of personal data, without the prior permission of the data provider should be allowed, unless it arises on account of legal obligation. This assumes importance as AVs may transmit and collect data from other vehicles and infrastructure; and
All entities dealing with data must have reasonable security standards, practices and policies in place that are commensurate with the data being collected.
“AI, you can drive my car!”
The increasing use and dependency on AI and advanced technologies in AVs make it important for these technologies to be fair, inclusive, and unbiased. Unlike humans, AI is not sentient and guided by assigned functions and programs. The range of error, bias, third-party interference, and miscalculation by AI may have dire consequences, ranging from mild injury to a disastrous fatality. As mentioned earlier, there have been road accidents by autonomous vehicles and the media reported[12] last year of an experimental passenger drone catching fire during a test flight.
A few ethical concerns also surface. In the event of two or more equally conflicting situations, say for example, life and death situations or a choice between death of a third party or passenger, what choice will be opted by the AI controlling these AVs? This ethical dilemma is popularly represented by the “trolley problem”, and gives rise to questions on moral responsibility of autonomous systems. More questions on how to prevent the use of AVs for illegal purposes, including terrorism, follow.
Hence, all these issues raise pertinent and concerning questions on the moral and ethical obligations of AI, which may include prioritization and preservation of human life, data protection, removal of bias, and fair decision-making process, based on ethical and established societal norms, amongst others.
On the subject of ethics of AI, nations across the world have come up with various research papers and recommendations, including UNESCO’s Recommendations on the Ethics of Artificial Intelligence[13], which was adopted by 193 member states on November 23, 2021. The Indian government too has taken initiatives to come up with a draft standard on “Fairness Assessment and Rating of Artificial Intelligence Systems”[14], which provides for detailed procedures for assessment and fairness certification for AI systems as well as a draft discussion paper on “Responsible AI for All: Adopting the Framework – A use case approach on Facial Recognition Technology”[15].
The road ahead!
It is obvious that one of the underlying themes of AVs is to provide reliable and safer mobility. But the journey has just begun. As noted by the RAND Corporation, “AVs will need to be driven hundreds of millions of miles and sometimes hundreds of billions of miles to prove their reliability in terms of fatalities and injuries.” [16]
The pressing need is to get the ball rolling for the widescale implementation of these mobility options. With greater power comes greater responsibilities, and the same is true in the AV sector. Lawmakers will need to address the glaring legal issues outlined above, apart from peripheral issues of urban planning, technology upgradation, infrastructure development, traffic management, resource and budget allocation, and workforce disruption.
To sum up, it is indisputable that for AVs to be a success and to make a difference vis-à-vis their intended purposes, they need to work where they are needed most, like in South Asia. The right foot forward in this transition is to 'Lead by Legislation', coming up with new laws which will pave the way for a sustainable, safer, and ethical future of mobility.
Endnotes
[1] SAE Levels of Driving Automation™ Refined for Clarity and International Audience; https://www.sae.org/blog/sae-j3016-update [2] https://time.com/6280480/asia-heat-wave-climate-april-2023/ [3] India, China, Indonesia, Pakistan and Bangladesh [4] https://www.worldbank.org/en/news/press-release/2022/12/14/urgent-action-needed-in-south-asia-to-curb-deadly-air-pollution [5] https://www.hindustantimes.com/videos/news/indian-navy-s-varuna-passenger-drone-unveiled-to-be-inducted-soon-watch-101664951644726.html [6] https://dronelife.com/2020/04/30/ehang-demonstrates-passenger-drones-for-emergency-response/ [7] Post amendment, it allows transfer of driving tasks to the vehicle, provided technologies used are in conformity with the United Nations regulations or can be overridden or switched off by the driver; https://unece.org/DAM/trans/conventn/crt1968e.pdf; https://perma.cc/UX89-K5TT [8] https://www.nhtsa.gov/sites/nhtsa.gov/files/2022-06/ADAS-L2-SGO-Report-June-2022.pdf; https://www.nhtsa.gov/sites/nhtsa.gov/files/2022-06/ADS-SGO-Report-June-2022.pdf [9] In Level 2 Advanced Driver Assistance Systems (ADAS), the automated system takes full control of the vehicle: accelerating, braking, and steering, requiring driver only to monitor and intervene upon failure of response by automated system. [10] https://www.ntsb.gov/investigations/AccidentReports/Reports/HAR1903.pdf [11] https://www.maricopacountyattorney.org/CivicAlerts.aspx?AID=751 [12] https://dronexl.co/2022/07/28/passenger-drone-evtol-catches-fire/ [13] https://unesdoc.unesco.org/ark:/48223/pf0000381137 [14]https://www.tec.gov.in/pdf/SDs/TEC%20Draft%20Standard%20for%20fairness%20assessment%20and%20rating%20of%20AI%20systems%20final%202022_12_27.pdf [15] https://www.niti.gov.in/sites/default/files/2022-11/Ai_for_All_2022_02112022_0.pdf [16] https://www.rand.org/content/dam/rand/pubs/research_reports/RR1400/RR1478/RAND_RR1478.pdf