Today, I'd like to talk Tesla. It's one of my favorite topics and something I've written about many times in this newsletter. I describe myself as both a Tesla enthusiast and a Tesla critic - a stance that keeps my eyes wide open to both the magic and the flaws of these vehicles. And yes, those flaws are wide enough to drive a Cybertruck through.
But let's start with the happy stuff.
On New Year's Day, my husband and I orchestrated our first synchronized "light show" in our driveway. We were giddy with excitement. While we've used this feature before on our individual cars, this was our first time creating a dual performance. It was like a double date with our vehicles, complete with both cars serenading us with Auld Lang Syne. In that moment, even the most hardened critic would have struggled not to smile.
I've owned my Tesla for two years now, and I still feel a spark of joy every time I back out of my driveway, accompanied by that signature spaceship sound. It makes me feel like I'm driving the future. I'll admit - I sometimes drive extra slowly just to savor that sound.
The whisper-quiet cabin, the glass roof, the concert-hall quality of my Spotify playlists through the ridiculous number of speakers (I play my music LOUD, like teenager LOUD), and yes, even the heated seats that pamper my California-softened self on our rare cold days - it all adds up to an experience that explains why Tesla owners love their cars so passionately.
But loving something means being honest about its flaws, especially when those flaws involve promises of technological revolution. This brings us to the heart of today's discussion: the divergent paths toward autonomous driving being pursued by Tesla and Waymo, and what they tell us about the future of transportation.
Two Roads to the Self-Driving Future
Tesla and Waymo represent fundamentally different philosophies in the pursuit of autonomous driving. Tesla, under Elon Musk's leadership, has committed to a vision-only approach, believing that cameras alone - combined with sophisticated neural networks - can achieve full autonomy. It's an elegant theory: humans drive using primarily visual input, so why shouldn't machines be able to do the same?
Waymo, on the other hand, has taken what might be called the "belt and suspenders" approach. As a subsidiary of Alphabet Inc., Waymo combines multiple sensor types - LiDAR, radar, and cameras - to create a comprehensive understanding of the vehicle's environment. Their vehicles map their surroundings with millimeter precision, creating a detailed digital twin of the world they navigate.
Understanding the Autonomous Vehicle Landscape
To truly grasp the magnitude of difference between Tesla and Waymo's approaches to autonomous driving, we need to first understand the established framework for measuring vehicle autonomy. The Society of Automotive Engineers (SAE) has defined six levels of driving automation, from Level 0 (fully manual) to Level 5 (fully autonomous). This framework helps us understand exactly where each company stands in their journey toward true self-driving capabilities.
As the visualization shows, Tesla's Full Self-Driving (FSD) system, despite its ambitious name, operates at Level 2 autonomy. This means that while the vehicle can control both steering and speed simultaneously, the driver must remain fully attentive and ready to take control at any moment. Think of it as having a talented but inexperienced student driver who needs constant supervision. Waymo, on the other hand, has achieved Level 4 autonomy within specific areas, allowing their vehicles to operate without human intervention in carefully mapped zones – more like having a skilled chauffeur who can handle most situations independently, but only along routes they know well.
But the differences between these companies go far deeper than their current autonomy levels. Their entire approaches to achieving self-driving capabilities represent two fundamentally different philosophies about how to solve one of technology's most complex challenges.
This interactive comparison reveals the stark contrasts in how these companies approach autonomous driving. Tesla's vision-only system, relying on cameras and neural networks, is like teaching a car to drive the same way humans learn – primarily through vision. Waymo's multi-sensor approach, combining cameras with LiDAR and radar, is more like giving the car multiple ways to perceive its environment, similar to how a pilot uses various instruments to fly a plane safely.
The deployment strategies of these companies are equally divergent, each with its own set of challenges and opportunities.
Tesla's approach favors rapid iteration and learning from a vast fleet of consumer vehicles, collecting data from millions of miles driven daily. This is like having millions of practice sessions happening simultaneously, each contributing to the system's overall learning. Waymo, conversely, takes a more controlled approach, thoroughly mapping and testing specific areas before expanding their service – more like methodically charting a new territory before allowing others to explore it.
The Promise vs. Reality
The contrast between these approaches becomes stark when we examine the timeline of promises versus achievements. Since 2015, Tesla's CEO Elon Musk has made bold predictions about the imminent arrival of full self-driving capabilities. Let's look at this timeline:
In December 2015, Musk predicted "complete autonomy" by 2018. In 2016, Tesla released a promotional FSD video that was later revealed in court to be staged. By 2017, Musk was promising that within two years, drivers would be able to sleep in their vehicles while they drove themselves. The predictions continued: a million robotaxis by 2020, "feature complete" FSD by the end of that same year, and Level 5 autonomy by the end of 2021.
As of the writing of this article in January 2025, Tesla FSD remains at SAE Level 2 autonomy.
Waymo, in contrast, has taken a more measured approach. Their autonomous vehicles operate in carefully mapped areas, with clear communication about their capabilities and limitations. While this has meant slower expansion, it has also resulted in a more reliable service. Today, Waymo operates fully autonomous taxis in select cities - no human safety driver required. Waymo’s vehicles are currently at Level 4- also known as "High Automation.” In SAE Level 4 vehicles can operate autonomously in specific conditions or environments, but humans can still take control. (The YouTube video included at the end shows these features.)
The Technical Reality
The core of this debate isn't just about timelines - it's about fundamental technical approaches. Tesla's vision-only system relies on neural networks trained on data from its vast fleet of vehicles. In theory, this should allow the system to learn from millions of real-world driving scenarios. However, this approach faces significant challenges in edge cases - rare but critical situations that can be difficult for AI to handle.
Waymo's multi-sensor approach provides redundancy and precision but at a higher cost and with more complex hardware requirements. Their LiDAR systems can create precise 3D maps of the environment, providing robust data even in poor visibility conditions. However, this approach requires extensive mapping and infrastructure, making rapid global scaling more challenging.
The Road Forward
As we move into 2025, both companies continue to refine their approaches, though with notably different strategies. Tesla has just rolled out FSD v13 to vehicles equipped with Hardware 4 (HW4) or AI4 computers in North America, while owners with Hardware 3 await an updated v12.6 version. This split rollout highlights one of the ongoing challenges in autonomous vehicle development: balancing hardware capabilities with software advancement.
The contrast between Tesla and Waymo's approaches becomes particularly stark when considering environmental challenges. Tesla's vision-only system, while innovative, faces fundamental limitations in conditions like heavy fog, severe rain, or snow - situations where cameras simply cannot gather reliable data. Waymo's multi-sensor approach, incorporating LiDAR and radar alongside cameras, provides redundancy that helps their vehicles navigate in these challenging conditions, though at a significantly higher cost per vehicle.
The question isn't just which approach will win - it's whether either approach alone is sufficient for truly safe, reliable autonomous driving in all conditions. Perhaps the future lies in a hybrid approach that combines the best elements of both strategies, finding a sweet spot between Tesla's scalable, lower-cost vision system and Waymo's more comprehensive but expensive sensor suite. Or perhaps we're still missing crucial pieces of the autonomous driving puzzle, pieces that neither company has fully identified yet.
The Human Element
As both a Tesla owner and a technology journalist, I find myself in a unique position to appreciate both the promise and the limitations of these systems. When my car negotiates a complex intersection flawlessly, it feels like the future has arrived. When it becomes confused by a simple lane merge, I'm reminded of how far we still have to go.
Final Thoughts
The autonomous driving landscape is evolving faster than my Tesla can back out of my driveway (which, by the way, has become significantly less nerve-wracking with recent updates). With FSD v13 now rolling out to HW4/AI4 vehicles across North America, we're seeing meaningful progress in Tesla's vision-only approach. The update brings improved object detection and smoother decision-making, while the holiday update adds practical features like rear cross-traffic alerts that make everyday driving safer. This last feature is particularly welcome - Tesla's previous rear visibility solution was essentially "trust your cameras and hope for the best," which led to more than a few close encounters with poles, walls, and that one particularly sneaky recycling bin.
Yet the race between Tesla and Waymo isn't just about who can release updates faster - it's about who can earn the public's trust while pushing the boundaries of what's possible. As we enter 2025, both companies are proving that there are multiple viable paths to autonomous driving. Tesla's democratized approach, pushing updates to hundreds of thousands of vehicles, contrasts with Waymo's methodical expansion of fully autonomous service areas. Both strategies have merit, and both face unique challenges.
As someone who experiences these technologies firsthand, I'm convinced that the future of autonomous driving will be shaped not by which company has the boldest vision or the most sensors, but by which one can consistently deliver on the promise of safer, more convenient transportation. For now, I'll keep enjoying my spaceship sounds and holiday light shows, while keeping my hands on the wheel and my eyes on the remarkable progress we're making toward a truly autonomous future.
I write about AI and technology twice weekly in my free newsletter, Deep Learning with the Wolf. Subscribe for more in-depth analysis of the latest developments in AI, autonomous vehicles, and other emerging technologies.
Frequently Asked Questions:
What's the current state of Tesla's Full Self-Driving (FSD)? As of early 2025, Tesla's FSD remains in beta testing, requiring active driver supervision. It's classified as a Level 2 autonomous system, meaning the driver must maintain control and attention at all times. Tesla plans to release FSD v13 and expand outside North America in Q1 2025.
Why doesn't Tesla use LiDAR like Waymo? Tesla believes that cameras alone, combined with advanced neural networks, can achieve full autonomy. Elon Musk has called LiDAR a "crutch" and argues that since humans can drive using primarily vision, vehicles should be able to do the same. This approach also keeps hardware costs lower and allows for easier mass production.
How does Waymo's approach differ from Tesla's? Waymo uses a multi-sensor approach combining LiDAR, radar, and cameras. They create detailed maps of their operating areas and focus on achieving full autonomy within specific geographic regions before expanding. Their vehicles operate at Level 4 autonomy within these areas, requiring no human intervention.
Can either system handle all weather conditions? Currently, both systems face challenges in severe weather. Camera-based systems can struggle with heavy rain, snow, or fog, while LiDAR can have difficulty with precipitation and road spray. Both companies continue to work on improving performance in adverse conditions.
What are the regulatory hurdles for full autonomy? Autonomous vehicles face varying regulations across different jurisdictions. Key challenges include liability frameworks, safety standards, testing requirements, and the need for new laws addressing fully autonomous vehicles. Companies must also navigate public trust and acceptance.
Is Tesla's vision-only approach viable for achieving full autonomy? This remains a subject of debate in the industry. While computer vision has made remarkable progress, many experts believe that sensor fusion (combining multiple types of sensors) provides necessary redundancy for safety-critical systems. The success of the vision-only approach may depend on further advances in AI and neural network technology.
What the WolfPack Is Watching:
Check out a world record Tesla light show in Finland. My brother asked: “What is the purpose of this feature again?” There is no ‘purpose.’ The cars play loud, synchronized music and do blinky-blinky lights back and forth at each other. It’s fun. And, even watching light shows on YouTube is fun—it’s like five minutes of brain candy.
What the WolfPack Is Reading:
TechCrunch. Waymo robotaxis are coming to Tokyo in 2025. (December 16, 2024.)
NotebookCheck. Tesla vs Waymo self-driving test has FSD 13 arriving at the destination way earlier. (December 6, 2024.)
NotATeslaApp. What’s Coming in Tesla FSD V13. (November 5, 2024.)
Tesla Oracle. Tesla FSD v12.6 is a scaled-down version of FSD v13 for HW3 owners (first impressions, Release Notes). (January 5, 2024.)
Non-Tesla and Waymo Related Reading:
Haydar Talib. Fear & Loathing in Voice Authentication - 10 Key Ideas on Deepfakes.
#autonomousdriving #selfdrivingcars #tesla #waymo #fsdv13 #autonomousvehicles #deeplearning #artificialintelligence #ai #futuretransport #robotaxis #computervision #lidar #sensortech #saelevel4 #machinelearning #autopilot #fsd #teslavswaymo
Share this post