POTHOLES.

 

GROK

Tesla FSD and Pothole Handling: Safety vs. EvasionTesla's Full Self-Driving (FSD) Supervised system, powered by its end-to-end neural network architecture, prioritizes safety above all else when dealing with road hazards like potholes. This means the system is programmed to avoid maneuvers that could introduce higher collision risks, even if it means driving straight over a pothole rather than swerving abruptly. Pothole detection itself relies on the vehicle's camera suite (Tesla Vision) to identify anomalies in the road surface, such as shadows, depth changes, or texture disruptions, but it's not foolproof—especially for subtle or flush potholes that blend into the pavement. The neural net weighs factors like vehicle speed, surrounding traffic, lane markings, and the pothole's severity to decide on a response.Programming for Safety vs. Evasion

• Safety-First Approach: FSD errs on the side of caution to prevent unintended consequences. For instance, if swerving left or right would require crossing into an adjacent lane with oncoming traffic (even if it's a few seconds away), the system defaults to staying in its path. This reduces the risk of side-swipes or loss of control, which data from Tesla's fleet shows are far more dangerous than minor wheel impacts. Elon Musk has emphasized this in past comments, noting that FSD's goal is to be "safer than a human driver" by avoiding high-variance actions. If evasion isn't deemed safe, the car might slow slightly (e.g., by 5-10 mph) to reduce impact force, but it won't brake hard unless the pothole is classified as a severe obstacle.
• Evasion When Feasible: If the pothole is detected early (typically 1-2 seconds out via forward-facing cameras) and the lane is clear, FSD will execute a gentle lateral shift—often just 1-2 feet—within the same lane to straddle or bypass it. This is more common on low-traffic roads or when the hazard is offset from the center. In crowded urban or highway scenarios, evasion drops in priority. Tesla's training data, crowdsourced from millions of miles of fleet driving, includes labeled pothole clips where the system learns to mimic human-like dodges only when probabilities align (e.g., <5% collision risk from the maneuver).
• Fallback Mechanisms:
  • Cloud Mapping: Tesla builds a "PotMap" by aggregating fleet data on reported rough roads. Vehicles download these micro-maps ahead of time, allowing proactive adjustments like raising adaptive suspension (on supported models like Model S/X) to soften impacts. This isn't full evasion but mitigates damage to low-profile tires and rims.
  • Driver Alerts: If you're not in FSD, a chime warns of upcoming hazards; in FSD, it logs the event for retraining.
  • Neural Net Evolution: Post-v12 (end-to-end AI shift), the system treats potholes more like debris, using the same object-avoidance logic as for tires or branches. However, it's not perfect—false positives (e.g., confusing road seams for potholes) can occur, leading to unnecessary hesitations.

In essence, the programming balances evasion with a conservative risk model: Evasion if low-risk, straight-line progression if safer overall. This has drawn criticism for "pothole-seeking" behavior in user reports, but Tesla argues it's statistically safer than aggressive swerves that could cause multi-vehicle incidents.Current Development Status (as of October 2025)FSD pothole handling has advanced significantly since Elon Musk's 2020 promise of "micro-maps" for avoidance, but it's still iterative and not at unsupervised autonomy levels. Here's a timeline based on recent updates:

FSD Version	Key Pothole Developments	Status Notes
Pre-v12 (Legacy, up to 2023)	Basic detection via radar/ultrasonic fusion; suspension adjustments only (no evasion). Frequent user disengagements for obvious potholes.	Limited to rough-road alerts; no active steering around hazards.
v12.x (2024)	End-to-end neural net enables first real-time evasions on clear roads. Fleet data starts building dynamic PotMap for predictive avoidance.	Improved detection rate (~70% for visible potholes), but still hits flush ones; users report 30-50% failure in urban areas.
v13.x (Early 2025)	Widespread evasion clips emerge (e.g., lane shifts on highways, dirt roads). Better at raised manholes and clusters of potholes.	Rolling out broadly; videos show successes like avoiding back-to-back potholes at 40-60 mph. However, ~20-40% hit rate in tests, especially without front bumper camera (HW3/AI3 vehicles).
v14.x (Mid-2025 onward)	10x AI parameter boost enhances nuance; proactive slowing/braking for rough patches. Reduced "nags" allow focus on hazards.	Imminent/implemented for many; early clips show assertive dodges (e.g., swerving around debris at 85 mph). Detection accuracy up to 85-90% per user anecdotes, but evasion still safety-gated. Tesla plans unsupervised FSD rollout by late 2025/early 2026, with potholes as a key training focus under their insurance umbrella.

As of now, v13.2.9 and v14 previews show FSD avoiding potholes in ~60-80% of documented cases, per community videos and reports, but it's not reliable enough for robotaxi fleets yet. Tesla continues fleet-wide data collection (e.g., via "report pothole" disengagements) to refine the model—every intervention trains the AI. For low-profile setups, users recommend 19"+ wheels with run-flats or vigilant supervision on poor roads. If you're experiencing frequent hits, enabling "Chill" mode or manual overrides for known bad spots helps while the system catches up. Overall, progress is rapid, driven by AI scaling, but full mastery of edge cases like these remains a work in progress.
While Tesla's Full Self-Driving (FSD) relies heavily on vision-based neural networks for pothole detection and conservative evasion (prioritizing safety over aggressive swerves), competitors like Waymo, Cruise, Zoox, and Mercedes-Benz (with Drive Pilot) use a mix of multi-sensor fusion, high-definition mapping, and vehicle-to-everything (V2X) communication. These systems aim to balance safety—avoiding secondary collisions—with evasion where feasible, but real-world incidents highlight ongoing challenges. Detection often leverages LiDAR for precise depth mapping (unlike Tesla's camera-only setup), enabling earlier identification of subtle hazards. However, evasion is gated by risk models similar to FSD: if swerving risks traffic interactions, the vehicle slows or proceeds straight. Development is iterative, with fleet data and simulations driving improvements, but none have achieved perfect reliability as of October 2025.

The Extent of the Pothole Problem in Major Cities: An OutlinePotholes—depressions in road surfaces caused by freeze-thaw cycles, water infiltration, heavy traffic, and poor maintenance—represent a widespread infrastructure challenge, exacerbated by aging roads, climate variability, and funding shortfalls. In major cities, they contribute to vehicle damage, accidents, traffic disruptions, and billions in economic losses. Globally, the issue is acute in urban areas with high traffic volumes and extreme weather, though severity varies by region. Below, I outline the problem's scope, focusing on statistics from 2023–2025, followed by comparisons across U.S. cities, states, and international contexts.1. Overall Extent in Major Cities

• Prevalence and Scale: Nearly 50% of major U.S. roads are in poor or mediocre condition, with cities bearing over two-thirds of national traffic and thus the brunt of pothole formation. An estimated 55 million potholes exist on U.S. roads (13+ per mile), concentrated in urban centers. In 2023–2024, pothole-related searches rose 107% from 2019 levels, peaking during winter-spring transitions. Internationally, 630,000 potholes were reported in England, Scotland, and Wales in 2023 alone—a five-year high.
• Economic Impact: Potholes cost U.S. drivers $28–$26.5 billion annually in vehicle repairs (up from $3 billion a decade ago), averaging $400–$600 per incident and affecting 15% of drivers yearly. Cities face repair costs of $35–$50 per pothole (plus $100–$150 mobilization), totaling millions—e.g., Chicago spent on 143,000 potholes in 2024. Globally, UK drivers lose £579 million ($750 million USD) yearly; broader infrastructure fixes could require $2.7 trillion in the U.S. over decades.
• Safety and Social Toll: Potholes cause loss of control equivalent to a 35 mph crash, contributing to accidents (up to 70% reducible with repairs), injuries (e.g., whiplash, fractures), and fatalities. In the U.S., 90% of drivers view them as a major safety risk; in the UK, they led to 643,318 breakdowns in 2024 (1,700+ daily). Climate change worsens this: Extreme heat, rain, and thaw cycles in Europe and Asia could increase potholes 20–25% post-storms, with southern/eastern Asia and western Europe most vulnerable.
• Trends: Reports surged in 2023–2025 due to heavy rains and deferred maintenance (e.g., COVID-era backlogs). Cities with apps/hotlines (e.g., Boston's 311) saw 10% drops in reports, but overall, 40% of U.S. roadways remain substandard.

2. U.S. Comparisons: Worst Cities and StatesRankings derive from Google search volumes for pothole complaints/repairs (2020–2025 data), reflecting perceived severity. Northern/midwestern cities dominate due to freeze-thaw cycles, but southern ones like Nashville surprise with traffic wear.

Rank	Worst Cities (U.S.)	Key Stats (2023–2025)	Worst States (U.S.)	Key Stats (2023–2025)
1	New York City, NY	Tops searches; fills 300,000+ yearly; $138M in settlements (2017–2023).	Washington	#1 overall; 3 cities in top 10 (Yakima #7, Spokane #6, Seattle #9); rainy winters amplify.
2	Los Angeles, CA	High urban traffic; 2nd in searches despite milder weather.	Minnesota	#2; Minneapolis #3 city; 1 in 5 drivers affected.
3	Minneapolis, MN	Severe winters; high per capita damage.	Michigan	#3; Detroit ranks high; aging infrastructure.
4	Nashville, TN	#1 in some 2022–2024 studies; rapid growth strains roads.	Tennessee	#9; Nashville leads complaints.
5	San Francisco, CA	Earthquake legacy + rain; top 10 searches.	Missouri	#14; St. Louis, Kansas City (#10 city) plagued.
6	Spokane, WA	Winter extremes; top 10 city.	Pennsylvania	#6; Philly in top 20 cities.
7	Yakima, WA	Rural-urban mix; #7 city.	New York	#10; NYC dominates urban complaints.
8	Kansas City, MO	#10 city; repairs avg. $460 (up 57% from 2021).	Illinois	Chicago: 143,000 potholes in 2024.
9	Seattle, WA	Frequent rain; #9 city; filled 23,000 in 2022.	Ohio	4 cities in top 50 (e.g., Cleveland #47).
10	Orlando, FL	Unexpected southern entry; humidity + traffic.	Georgia	Atlanta reports rising; heat cracks roads.

• Insights: Northeast/Midwest lead (24% more potholes per capita than average); e.g., 1 in 5 Northeast drivers repair annually. Southern cities like Orlando/Nashville rank high from construction/traffic, not just weather. Best states: Nevada, Wyoming, Alabama (fewest complaints).

3. International ComparisonsPotholes are global but vary by climate, funding, and maintenance. Europe faces "pothole plagues" from wet winters; Asia sees monsoon spikes. U.S. costs dwarf others per capita due to vehicle reliance, but repair backlogs are similar.

Region/Country	Key Cities	Extent & Stats (2023–2025)	Comparison to U.S.
UK (Europe)	London, Stoke-on-Trent	630K reports (2023); RAC: 6,575 Q2 2025 breakdowns (9% rise); £579M driver costs. Stoke tops searches.	Similar scale (1.7x more breakdowns since 2006 vs. U.S. 107% search rise); £5B gov't fund to 2025 mirrors U.S. Infrastructure Act. Worse per mile due to denser traffic.
Sweden (Europe)	Stockholm, Växjö	National Pothole Week (2025); SEK 19.1B maintenance (2023); apps boost reporting.	Fewer overall (better funding); U.S. has 55M potholes vs. Sweden's targeted fixes. Climate worsens both equally.
Other Europe	Amsterdam (NL), Marseille (FR)	Harmonized data: 120K sensors track traffic/potholes; heavy rain up 25% post-2023.	Top road quality (e.g., Netherlands #1 globally); 40% fewer potholes with drainage vs. U.S. Northeast's 24% excess.
Asia (e.g., Philippines)	Manila	Monsoon potholes on major roads (2024); floods + rain create hazards.	Acute seasonal spikes (20% post-storms like U.S. winters); less data, but higher fatality risk (WHO: Asia tops road deaths). Poorer infrastructure amplifies vs. U.S. urban fixes.
Australia	Darwin	#1 per capita searches (3.56/1,000 people); aligns with population density.	Milder issue (fewer freeze cycles); U.S. cities like NYC have 10x higher complaints per capita.

• Global Insights: U.S. leads in total costs ($28B/year) due to scale, but Europe (e.g., UK) has higher per-driver breakdowns. Asia's monsoons mirror U.S. winters but with less funding (e.g., Manila's reactive fixes). High-quality roads (e.g., Netherlands, Singapore) reduce potholes 40–60% via investment; climate models predict 20%+ rises everywhere by 2050.

4. Mitigation and OutlookCities like Boston (10% report drop via apps) and Baltimore (19K repairs in 90 days) show progress with tech/crowdsourcing. U.S. Infrastructure Act funds $1.57M/mile for poor roads; UK/ Sweden allocate billions. Yet, without sustained investment, potholes could cost $185B/year U.S.-wide. For drivers: Report via 311/hotlines, inflate tires, and claim reimbursements where possible. The problem is solvable but demands proactive infrastructure over patchwork.