Washington Wants to Kill 50 Robotaxi Rules

Uber often allowed competitors like Lyft to enter new markets first, test the regulatory waters, and establish legality before making its own entry. This strategy allowed Uber to learn from its rivals' experiences and refine its market penetration.

Waymo currently serves a similar role for other AV companies, acting as a catalyst by navigating and testing complex regulatory environments. The push for a federal framework aims to unify 50 disparate state regulations into a single, cohesive standard, which proponents believe will accelerate the broader adoption and scaling of autonomous vehicle technology across the nation.

Moving too slowly with regulations can kill innovation, preventing life-saving technologies from reaching the public. Conversely, moving too fast raises serious concerns about potential public safety hazards and unintended consequences.

The panel discussing these issues argued that delayed regulation itself can result in preventable human casualties, underscoring the urgency of finding an optimal pace. The challenge lies in creating a framework that encourages rapid technological advancement while maintaining rigorous safety standards.

This decision will significantly impact Tesla's expansion plans across the continent. Currently, FSD is approved and operating in five European Union countries: the Netherlands, Denmark, Belgium, Estonia, and Lithuania.

To counter criticisms regarding a lack of transparency and safety concerns, Tesla has shared comprehensive safety data with European regulators. This data aims to demonstrate the system's robustness and safety performance, hoping to secure broader approval and adoption across the EU.

Statistical data from European trials indicates significant safety benefits for Tesla's Full Self-Driving (FSD) system. FSD is currently approved in five European countries and is used by over 70,000 customers, covering more than 1 million kilometers per day. Performance data measured across 100 million kilometers on public EU roads shows that FSD is four times less likely to result in a crash compared to manual driving.

This is particularly relevant given that European roads record approximately 19,400 annual deaths, with human error being the primary contributing factor. The data suggests that FSD's deployment could lead to a substantial reduction in accidents and fatalities by mitigating risks associated with human drivers.

This achievement serves as a testament to the effectiveness and rapid adaptability of Tesla's neural networks, which can quickly improve and enhance safety in new environments.

Full Self-Driving (FSD) benefits drivers most significantly in highway environments, where complacency often leads to human error. The system's ability to maintain vigilance and react consistently provides a distinct safety advantage. Globally, the total cumulative driving miles with FSD have now exceeded 14 billion, indicating widespread usage and continuous data collection for further improvements.

Major incidents were reduced by 45% on highways, 69% on arterial roads, 50% in city environments, and 40% on local roads. The system also demonstrated remarkable improvements in emergency braking activations, reducing them by up to 84%.

Harsher braking incidents decreased by up to 81%, reflecting smoother and more predictive driving. Furthermore, minor collisions, which often go unreported, were effectively reduced by an impressive 99%, highlighting the system's ability to mitigate a wide range of road incidents.

Many minor collisions, such as fender benders or scrapes, are resolved privately or simply overlooked without official reports. Tesla's reported 99% reduction in minor collisions is therefore highly significant.

It is suggested that this figure could actually represent an even greater reduction, possibly 99.99%, when compared to the true, untracked total volume of minor collisions. These figures reflect data collected across North America, demonstrating the system's broad impact on reducing a category of incidents that often escapes official statistics.

The intentional 2% failure rate is not a flaw but a crucial part of the development process, designed to identify challenging edge cases and drive further growth. By allowing for a non-100% success rate, Tesla encourages its system to encounter and learn from difficult scenarios rather than gaming the system by only testing easy drives.

Targeting an artificial 100% success rate would discourage the testing of challenging scenarios, which are essential for developing a truly safe and capable autonomous system. This approach ensures that the system continuously improves by addressing its weaknesses.

Brian White referenced an instance at Washington Mutual where a 0.5% default rate was cheered, but the CEO deemed it too low, suggesting they were not issuing enough loans to enough people and thus under-performing. This perspective highlights that a perfect score might indicate a lack of ambition or insufficient exploration of challenging boundaries.

Tesla's testing methodology similarly focuses on deliberately finding failure points to understand and solve them, rather than simply optimizing for a perfect test score. This approach ensures the system is rigorously challenged and continuously improves its performance by addressing weaknesses in real-world conditions, ultimately leading to greater overall safety and capability.

The system also maintains strong performance in fog, dispelling the common misconception that autonomous systems would struggle in such conditions.

This consistent performance is observed regardless of the time of day, further proving the system's robustness. Interestingly, human driver behavior during inclement weather often normalizes safety metrics, as drivers tend to be more cautious, which helps FSD's relative safety performance in challenging conditions.

In countries where FSD is approved, data shows a four times fewer collision rate. This substantial reduction underscores the system's enhanced ability to prevent accidents and improve overall road safety.

Further metrics highlight these advancements: FSD vehicles exhibit nearly 14 times fewer automatic emergency braking (AEB) activations and 10 times less harsh acceleration. This indicates a smoother, more proactive driving style that avoids sudden, reactive maneuvers.

Additionally, there are eight times fewer harsh braking events and hard swerves, showcasing the system's predictive capabilities and stable operation. These combined improvements suggest FSD offers a significantly safer driving experience than human-operated vehicles.

The country currently permits Tesla to conduct limited FSD tests with trained drivers on selected routes, indicating a measured approach to integrating autonomous technology. The initiative stems from a desire to address transportation challenges faced by elderly populations.

One municipality has expressed interest in becoming a pilot site for future robotaxis, aiming to provide seniors with greater access to independent transportation. Notably, the recent IF Metall labor strike, which affected some Tesla operations in Sweden, did not impact the approval process or consideration of these FSD pilot programs, suggesting a clear distinction in regulatory discussions.

Of this impressive total, approximately 5.4 billion miles were accumulated in city driving, which represents some of the most complex and challenging autonomous driving environments. This substantial mileage provides a vast dataset for continuous improvement and validation of the FSD system.

According to Tesla's own safety hub, FSD drives with a collision rate that is seven times lower than that of human drivers. This statistic underscores the safety advantages of the FSD system, reinforcing its potential to make roads safer as adoption grows. The continuous accumulation of miles further solidifies Tesla's leadership in autonomous driving data collection and real-world testing.

It took Tesla approximately 3.5 years to accumulate its first billion miles of FSD data, a testament to the initial stages of development and gradual rollout. However, the pace has exponentially increased, with the final billion miles being reached in a mere 28 days.

This rapid acceleration indicates a significant increase in user satisfaction and confidence, leading to broader adoption among Tesla owners. The exponential growth in mileage accumulation provides an invaluable feedback loop for Tesla's AI and engineering teams, enabling faster improvements and refinements to the FSD system. This trend is crucial for scaling the technology and achieving full autonomy.

This expansion significantly boosts Waymo's presence, bringing its total coverage to 14 cities across the United States. The introduction of paid services in these cities marks a crucial step in commercializing Waymo's autonomous technology and growing its operational footprint.

Despite this geographical expansion, Waymo's fleet size remains relatively modest, staying under 4,000 vehicles. This indicates a controlled and gradual rollout, prioritizing safety and system robustness over rapid fleet scaling. The addition of new markets allows Waymo to gather diverse operational data and refine its service in varied urban environments.

The high upfront costs for purchasing and equipping autonomous vehicles create substantial cash flow pressures for operators, even for companies backed by deep-pocketed entities like Alphabet. These financial demands can hinder rapid expansion and pose risks to profitability in the short term.

While the long-term potential of robotaxis is significant, the immediate future involves considerable investment and operational expenses. This creates a "valley of cash flow crunch" where companies must navigate substantial expenditures before achieving widespread profitability. The need to pay for these advanced vehicles puts a significant short-term downside potential on their expansion efforts.

Custom vehicles, such as those developed by Zoox, are inherently expensive to design, manufacture, and deploy due to their bespoke nature. This high unit cost can severely limit the scalability and profitability of autonomous fleets.

Conversely, utilizing off-the-shelf cars, which are mass-produced and widely available, can reduce unit costs by as much as 50%. This substantial cost saving makes scaling a robotaxi fleet much more financially viable. Furthermore, mass-produced vehicles often offer more predictable reliability and maintenance profiles compared to hand-built, specialized vehicles, contributing to operational efficiency.

Waymo currently holds a significant lead in actual paid, unsupervised robotaxi miles, where there is no human safety driver. Tesla's FSD Supervised system, by contrast, still requires a person to be present and attentive, ready to intervene if necessary.

For Tesla to truly catch up and compete effectively in the robotaxi market, it needs to transition from supervised FSD to real, paid, unsupervised autonomous miles. The next critical hurdle for Tesla's scalability in this segment will be demonstrating the capability to operate fully autonomously without human oversight and rapidly expanding its fleet size to offer competitive services.