Three companies. Three fundamentally different answers to the same question: what should a robotaxi actually be? Waymo, Zoox and Tesla are each pursuing commercial autonomous ride-hailing, and each has made a distinctive set of choices about the vehicle — choices that reflect different assumptions about what matters most in a robotaxi, what trade-offs are acceptable and what the long-term economics of the business require. Those choices are not merely technical curiosities. They carry significant implications for safety certification, regulatory pathway, passenger experience and the shape of the market that eventually reaches Australian cities. Understanding the differences is increasingly important for anyone — regulators, urban planners, transport researchers or future passengers — who wants to think clearly about what autonomous ride-hailing will actually look like when it arrives.
Three Companies, Three Philosophies
At the broadest level, the three approaches can be characterised as follows. Waymo’s philosophy is to take a proven high-quality production vehicle and rebuild it comprehensively for autonomous operation, adding its own sensor hardware, computing systems and safety architecture while retaining the structural integrity and crash safety ratings of an established platform. Zoox’s philosophy is to discard the conventional vehicle form entirely and build something new from a clean sheet, optimised from the outset for shared autonomous taxi operation. Tesla’s philosophy is that the autonomous capability — specifically, a vision-based artificial intelligence system — is the defining innovation, and that the vehicle itself can evolve from Tesla’s existing consumer platform without fundamental structural reinvention. Each philosophy is internally consistent. Each carries consequences.
Waymo’s Approach: Modify the Proven
Waymo’s current commercial fleet uses a modified Jaguar i-Pace, a premium electric SUV that brings a high baseline of passive safety, structural rigour and quality control from an established automotive manufacturer. To this foundation, Waymo adds its own sensor suite — lidar, radar and cameras working in combination — along with purpose-built computing hardware and the software system that has now accumulated more than 170 million fully autonomous miles. The advantage of this approach is that Waymo begins with a vehicle that already meets stringent automotive safety standards. The modifications layer autonomous capability on top of a proven foundation rather than requiring the safety case to be built from scratch. Waymo’s published safety data reflects the outcome of this approach: a dramatic reduction in serious injury incidents compared to human-driven vehicles across comparable populations. The limitation is that a vehicle designed for conventional driving carries inherent compromises when repurposed as a shared taxi — the interior layout, seating configuration and user experience are constrained by the original design intent.
Tesla’s Approach: Software-First, Consumer DNA
Tesla’s entry into purpose-specific robotaxi territory comes via the Cybercab — a two-seat vehicle designed without a steering wheel or conventional driver controls. The Cybercab carries forward the fundamental design philosophy that has characterised Tesla’s autonomous vehicle programme: vision-based artificial intelligence, drawing on camera inputs without lidar or radar, as the primary perception system. This approach, if it works as intended, has compelling economic advantages — cameras are significantly less expensive than lidar systems and can be integrated more cleanly into a vehicle’s body design. The trade-off is that vision-only perception has historically been considered more vulnerable to edge cases — unusual lighting conditions, obscured lane markings, unexpected objects — than multi-sensor systems. The technical debate between vision-only and multi-sensor approaches is ongoing, and neither camp has conclusively resolved it. The Cybercab’s two-seat configuration also limits shared-ride economics compared to four-seat platforms, which has implications for the cost per mile that operators can offer passengers.
Zoox’s Approach: Designed From the Ground Up
Zoox’s vehicle represents the most radical departure from conventional automotive design among the three. Built from a clean sheet with no steering wheel, no driver’s seat and a bidirectional drivetrain that allows the vehicle to travel in either direction without turning around, the Zoox robotaxi configures its four-seat interior around a face-to-face passenger arrangement designed for shared rides. Every aspect of the vehicle — sensor placement, weight distribution, power architecture, interior volume — is optimised for the specific operating requirements of a fleet taxi rather than adapted from a product designed for individual consumers. The advantage of this approach is the elimination of inherited compromises: the sensors can be positioned exactly where they work best, the interior can be configured exactly as passengers need it and the vehicle’s operational characteristics can be tuned precisely to urban taxi use. The challenge is that every element of the safety case must be demonstrated from scratch, without the benefit of an established vehicle platform with an existing crash safety record.
Sensor Philosophy: The Deepest Technical Divide
Beneath the visible design differences lies a more fundamental technical divergence: the question of how an autonomous vehicle perceives the world around it. Waymo employs a multi-sensor approach combining lidar — which uses laser pulses to create detailed three-dimensional maps of the vehicle’s environment — with radar and cameras. Each sensor type has different strengths: lidar provides precise spatial geometry; radar is effective in poor visibility and can detect velocity directly; cameras provide the rich colour and texture information that humans rely on for visual recognition. Zoox uses a similar multi-sensor philosophy. Tesla’s Cybercab relies on cameras alone, without lidar or radar, arguing that human-level visual intelligence — processing camera images with sufficiently sophisticated AI — is sufficient for safe autonomous operation. The development of standardised safety benchmarks like Waymo’s Reference Driver model will eventually provide tools for comparing these approaches on a common basis, which matters significantly for how regulators assess each platform.
What Each Approach Means for Australian Roads
Australian roads present a specific set of challenges for autonomous vehicles: left-hand drive, diverse urban environments ranging from dense inner-city grids to sprawling suburban arterials, unpredictable wildlife on roads in regional areas and a regulatory framework still being finalised by the National Transport Commission. The Waymo approach — modifying established vehicles — creates a relatively tractable regulatory pathway because the base vehicle already meets recognised standards. The Zoox approach requires building a new certification case but offers design flexibility that may ultimately produce a better urban taxi product. Tesla’s vision-only approach faces the most uncertainty in markets with unfamiliar road marking conventions or lighting conditions not well represented in US training data. None of the three approaches is obviously best suited to be first to Australian streets — the outcome will depend as much on regulatory readiness and commercial strategy as on technical architecture.
What This Means for the Future Rider
For the Australian who will eventually hail an autonomous taxi, the design philosophy of the vehicle matters in practical terms. A Waymo ride today feels familiar — it is, after all, a Jaguar SUV — with the notable absence of a driver. First-time riders consistently describe the experience as simultaneously mundane and remarkable. A Zoox ride would be structurally different: a purpose-built interior with face-to-face seating, more like a small private cabin than a taxi. A Tesla Cybercab would be intimate and compact, with a two-seat configuration that suits point-to-point travel but limits shared-ride scenarios. The question of which of these experiences Australian riders will encounter first remains open — but the fact that three substantively different answers to the question “what is a robotaxi?” are now all in commercial operation makes the Australian arrival of this technology feel considerably less theoretical than it did even two years ago.
Sources
- Zoox — The Purpose-Built Robotaxi Platform (zoox.com)
- Waymo — The Waymo Driver Technology Overview (waymo.com)
- Tesla — Cybercab Autonomous Vehicle (tesla.com)
- Waymo — Published Safety Performance Data (waymo.com)
- National Transport Commission — Automated Vehicle Program (ntc.gov.au)
- Featured image: Aerial view of complex urban infrastructure via Pexels
