[AutonomousVehicles] The $400 Billion Robo Taxi Race is on

[Cornelius Butler]

There is an excellent comprehensive overview of the current status of the
Robo Taxi Race in the U.S. Here is the video link and summary.

Video Link: https://www.youtube.com/watch?v=TxLUXkTjVZg

Video Summary: The $400 Billion Dollar Robotaxi Race Is On | Tesla VS
Google & Amazon

Channel: Ryan Shaw
Host: Ryan Shaw
Published: August 8, 2026

SAE Taxonomy, Legal Architecture, and Self-Certification

SAE J3016 & ISO 22736 Standards: Automated driving levels 0 through 5
originated from SAE International recommended practices and were later
co-developed with ISO. The levels define operational roles and legal
responsibility rather than directly testing vehicle capability [01:47].

Self-Certification vs. Type Approval: Unlike Europe’s type-approval system,
U.S. Federal Motor Vehicle Safety Standards (FMVSS) generally rely on
manufacturer self-certification rather than federal pre-sale performance
testing. NHTSA recently launched a $5 million consortium with SAE to
develop dedicated autonomous-vehicle performance standards [04:23].

Legal Implications of Automation Levels: A vehicle’s declared SAE level can
affect which state permits apply and who carries legal responsibility
during a crash [01:13]. In federal filings, Tesla classifies Autopilot and
FSD Supervised as Level 2 driver-assistance systems, meaning the human
driver remains legally responsible [03:57].

Breakdown of Levels 0 Through 5

Levels 0–2: Driver Support

Level 0: Provides momentary warnings or interventions, such as Automatic
Emergency Braking or Lane Departure Warning [05:10].

Level 1: Continuously controls either steering or speed, but not both at
the same time. Adaptive Cruise Control is a common example [05:28].

Level 2: Continuously controls both steering and speed. Examples include
Tesla FSD Supervised, Ford BlueCruise, and GM Super Cruise [05:34]. The
human driver must continuously monitor conditions and remains responsible
for the vehicle [05:47].

Level 3: Conditional Automation

The system legally performs the driving task within a limited Operational
Design Domain, or ODD. Drivers may take their eyes off the road but must be
prepared to take control within seconds if the system requests intervention
[09:41].

Level 4: High Automation

The vehicle performs the complete dynamic driving task within defined
geographic, weather, speed, or roadway conditions. No human driver or
takeover fallback is required while operating within those conditions.
Commercial robotaxis generally fall into this category [14:52].

Level 5: Full Automation

The vehicle could theoretically operate anywhere a human could drive, under
essentially all roadway and weather conditions, without geofences or
mapping restrictions. Level 5 is not commercially available today [16:05].

Level 3 Commercial Hurdles and Automaker Retreats

Early Level 3 Systems: Honda introduced an early commercial Level 3 system
in Japan through the limited-production Legend sedan with Traffic Jam
Pilot. Mercedes-Benz later introduced Drive Pilot in Germany, California,
and Nevada [10:20].

Automaker Liability: When a Level 3 system is legally operating and causes
a collision, responsibility can shift from the vehicle owner toward the
manufacturer [11:46].

Commercial Retreat: Mercedes-Benz and BMW have recently pulled back from
consumer Level 3 systems on flagship vehicles. Challenges include expensive
LiDAR and other hardware, narrow operating conditions, speed restrictions,
subscription costs, and increased manufacturer liability. Both companies
have increasingly focused on advanced hands-off, eyes-on Level 2 systems
instead [12:30].

Ford’s Alternative Approach: Ford plans to introduce an internally
developed eyes-off Level 3 version of BlueCruise by 2028 on mass-market EV
platforms, with the goal of reducing system costs by roughly 30% [14:27].

Level 4 Fleet Operations and Remote Assistance

Operational Design Domain: Level 4 systems operate within specific limits
covering geography, speed, weather, road types, and other conditions.
Waymo, for example, has gradually expanded from surface streets to more
complex environments, including freeways [15:15].

Remote Driving vs. Remote Assistance:

Remote Driving, or Teleoperation: A remote human directly controls
functions such as steering and pedals, effectively transferring control to
a human operator [17:26].

Remote Assistance, or Fleet Response: A remote operator provides contextual
information to the autonomous system, such as helping interpret unusual
construction cones, while the vehicle itself continues controlling its
movement. This can preserve the vehicle’s Level 4 classification [17:32].

Autonomous Freight: Aurora Innovation has accumulated nearly 440,000
driverless commercial Class 8 trucking miles across 10 Sunbelt routes and
is targeting a production run-rate of 1,000 driverless trucks per year by
late 2026 [18:34].

Regulatory Permitting Landscape and Crash Repercussions

California Permitting Hierarchy:

Safety-Driver Testing: 27 companies are approved, including Tesla, Waymo,
Zoox, and NVIDIA [19:44].

Driverless Testing: Six companies are approved: Waymo, Zoox, Nuro, WeRide,
Tensor, and AutoX [20:06].

Commercial Deployment: Three companies are approved: Waymo, Nuro, and
Mercedes-Benz [20:21].

Tesla’s California Classification: Tesla operates under a standard
Charter-Party Carrier permit requiring safety drivers because its system is
registered as Level 2 [20:46].

Federal Exemptions for Zoox: NHTSA granted Zoox a two-year commercial FMVSS
exemption covering up to 2,500 purpose-built, bidirectional robotaxis
without traditional steering wheels or pedals [21:57].

The Cruise Precedent: General Motors’ Cruise division faced permit
revocation and a $500,000 criminal penalty following an undisclosed
pedestrian-dragging incident in late 2023. GM ultimately shut down its
dedicated Cruise robotaxi operation after cumulative losses exceeding $10
billion [23:06].

Safety Data, Studies, and Scrutiny

Waymo Safety Benchmarks: Across more than 220 million driverless miles,
Waymo reports 82% fewer injury crashes and 93% fewer pedestrian injury
collisions compared with human-driving benchmarks [25:46].

IIHS Independent Validation: An IIHS analysis covering approximately 50
million Waymo driverless miles reportedly found an overall crash rate 68%
below human drivers. The reductions were 76% in Phoenix, 71% in Los
Angeles, and 35% in San Francisco. Austin showed a slight 4% increase,
although the sample there was considerably smaller [27:17].

Tesla FSD Safety Reporting: Tesla has moved from primarily reporting
highway Autopilot statistics to evaluating FSD across more than 1.3 billion
miles and comparing the results with Tesla’s manually driven fleet. Tesla
reports approximately seven times fewer major collisions, using airbag
deployments under FMVSS definitions and a five-second post-disengagement
crash-attribution window [29:45].

Congressional and Academic Scrutiny: U.S. senators and independent safety
researchers have questioned factors including differences in vehicle age
between newer Teslas and the approximately 12-year-old average U.S. vehicle
fleet, Tesla’s collision definitions compared with federal tow-away
standards, and potential telematics reporting gaps following severe crashes
[31:53].

Insurance Disruption and the Economics of Autonomy

Actuarial Risk Pricing: Insurers such as Lemonade have introduced policies
offering approximately 50% discounts for miles driven using Tesla FSD,
based on real-time API telematics. Tesla Insurance also assigns its highest
safety-score treatment while FSD is operating [37:54].

Sensor Repair Costs: Autonomous-driving technology may reduce crash
frequency, but the savings can be partially offset by higher repair
expenses for cameras, radar, sensors, and required calibration [38:46].

Long-Term Economics: The robotaxi market is likely to increasingly focus on
total cost per mile. As autonomous fleets expand, accident liability and
insurance risk could shift from individual drivers toward the companies
operating autonomous fleets [39:15].
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://nfbnet.org/pipermail/autonomousvehicles_nfbnet.org/attachments/20260815/516645aa/attachment.htm>