Tesla's Robotaxi Faces Numerous Issues! A Beta Version on the Move, Trust at a Standstill - Austin's "Test Drive of the Future" Report

Introduction: The Stumbling of the Promised "Future"

"2025 will be the year of fully autonomous driving." CEO Elon Musk has repeated this claim over the past few years. However, Tesla's robotaxi "Cybercab" (vehicle base is Model Y), which began public rides on June 22 in Austin, Texas, managed to accumulate a record of "failures" in just 72 hours. These included driving in the wrong lane, sudden phantom braking, drop-off instructions in the middle of intersections, and mounting curbs... The Verge's video compilation titled "All Mistakes List" confronts the reality of the "dream mobility experience."theverge.com

1. Background to the Launch: Repeated Delays and the "August 8" Promise

• April 2019 ― Musk declared "1 million robotaxis by the end of next year."

• April 5, 2023 ― After reports of "cancelling Model 2 to focus on robotaxis," Tesla announced an "August 8 event."twitter.com

• June 22, 2025 ― Paid service began with 10-20 vehicles (in-app payment: flat rate of $4.20).

This "pre-release" was based on the software version FSD 12.8 series, renamed "Supervised Full Self-Driving," applied to vehicles with the steering wheel removed. An employee called a "safety monitor" accompanies the ride, and operational data is transmitted in real-time to Tesla's headquarters data center.

2. Details of Incidents (First Day to 72 Hours)

The estimated number of rides recorded in the same 72 hours is 1,200 to 1,500. This means that **approximately once every 100 to 120 rides, a "glitch exposed online without safety monitor intervention"** occurs.

3. Temperature on Social Media: Praise, Ridicule, and Investors' Prayers

• Positive Side

  • Tesla influencer Sawyer Merritt praised it, saying, "20 rides in 36 hours, 92 miles traveled, zero interventions."x.com

  • On X (formerly Twitter), #SmoothRide briefly trended.

• Negative Side

  • On Reddit/r SelfDrivingCars, there were numerous harsh criticisms, saying it was "just forcibly porting FSD to 'steering wheel-less'."reddit.com

  • On TikTok, influencers posted reenactment videos of sudden braking as a "jerk challenge."

• Neutral/Ironic Side

  • The $4.20 fare became a "420 joke" meme. Daily Galaxy delved into the intention behind the pricing.dailygalaxy.com
    
    

4. Technical Issues: Camera-Dependent FSD and the Trap of "Shadows and Reflections"

1. Sensor Configuration

   • Tesla: 8 cameras + radar abolished. No use of LiDAR or HD maps.

   • Competitor Waymo: LiDAR + radar + cameras + HD maps.

2. Mechanism of Phantom Braking

   • Mistakes shadows and road markings for pedestrians or obstacles, triggering AEB (Automatic Emergency Braking).

   • An internal leak in May 2023 recorded **over 1,500 "sudden braking complaints"** in the database.theverge.com

3. Bias in Learning Data

   • Data set centered on the Sunbelt area → Lacks examples of rainy weather, nighttime, and snowy roads.

   • Tesla integrated a "snow model" by the end of 2024, but it is still in the beta stage.

5. Regulation and Administration: Federal Void and Texas's "Lenient Licensing"

• NHTSA (National Highway Traffic Safety Administration)

  • On June 23, following online videos, Tesla was requested to provide information. The investigation is at the "preliminary inquiry" stage.reuters.com

• Texas Transportation Commission

  • The new AV permit system, enacted in April, added "permit revocation in case of significant risk" to its clauses. Still, it remains a post-reporting system.

• Regulatory Vacuum

  • The federal government does not adopt a pre-approval system, and from defect recognition to recall, it takes at least several months.

  • The large-scale Cruise recall in 2023 (San Francisco incident) also resulted in "operations being halted only after a roadblock accident."

6. Business Impact: Stock Prices Rise, but Unexpected "Recall Costs"

The stock price was bought simultaneously with the announcement, butif a formal recall occurs, the average repair cost per vehicle is estimated at $3,000.For a scale of 100,000 vehicles, it would be 300 million USD. No investor overlooks such risks.

7. Lessons from