The Robocar features over 100 integrated sensors, including 37 cameras, 5 lidars, and 11 radars, to ensure unmatched visibility and safety/ Image: Tensor
At the upcoming Dubai World Congress for Self-Driving Transport 2025, Tensor is set to debut the world’s first personally owned Level 4 autonomous vehicle, the Tensor Robocar. This groundbreaking vehicle, designed from the ground up for autonomy and privacy, represents a major leap in personal mobility, combining advanced AI, industry-leading safety, and innovative features to empower individual ownership in a way no other autonomous car has before. Customer deliveries are projected in H2 2026.
The debut of Tensor Robocar in Dubai:
Tensor, a Silicon Valley-based AI company, is introducing its Robocar publicly in the Middle East for the first time at the Dubai World Congress for Self-Driving Transport, held on September 24–25, 2025, at the Dubai World Trade Centre. The event, organized by the Roads & Transport Authority (RTA) – Government of Dubai, expects over 3,000 attendees and more than 85 speakers focused on AI-powered mobility, smart infrastructure, safety and sustainable transport. For Tensor this is strategic and symbolic. The company calls the Robocar a category-defining product and says the Dubai showcase marks a defining milestone in its mission to “transform transportation.” As Amy Luca, Tensor’s Chief Marketing Officer, puts it: “When the world shifts, how will you move? … This isn’t a concept—it is the world’s first real personal Robocar. It’s time to Own Your Autonomy.”
What the Tensor Robocar actually is
Tensor is not a converted EV: it is engineered from the ground up for autonomy. The design and systems were built on “a decade of proprietary engineering” and native support for autonomy rather than being an add-on. Customer deliveries are expected in the second half of 2026, with Dubai positioned at the forefront of this new mobility era.Key hardware and architecture highlights:
- Sensors: Over 100 seamlessly integrated sensors, including 37 cameras, 5 lidars, and 11 radars to deliver uncompromised visibility.
- Materials & sightlines: Radar-transparent materials, unobstructed lidar sightlines, and a low hood profile to prioritize perception and safety.
- Mechanical reliability: Washer systems for lidar and cameras plus intelligent sensor covers and protective sensor covers — features designed specifically for sand, dust and desert climates.
- Redundant electrical / electronic architecture: Full-stack redundancy across sensors, compute, communications, drive-by-wire, power, control and thermal management to achieve fail-operational performance in any scenario.
- Compute & partners: An onboard supercomputer provided by Nvidia (8,000 TOPS); Sony supplies 17-MP automotive-grade cameras; manufacturing is supported by VinFast; cloud and enterprise support includes Oracle; Marsh is partnering on insurance. Tensor also lists leading semiconductor partners across its supply chain.
- User interface & ergonomics: Dual Mode capability — drive or be driven, plus interior innovations like a foldable steering wheel and a sliding display.
AI stack and operation:
- The vehicle is powered by the Tensor Foundation Model, a Transformer-based architecture trained on vast real-world and simulated datasets with advanced sensor fusion.
- Tensor’s perception, prediction and planning are learned rather than hardcoded; the stack is built to handle desert glare, dust, night, fog, and heavy rain.
- The AI runs as two complementary systems modeled on human cognition: System 1 for fast, reflexive reactions and System 2, a multimodal Visual Language Model, for higher-order reasoning in rare and complex edge cases.
Operational features for private ownership:
- Intelligent sensor-cleaning and protective covers for sand/dust resilience.
- Autonomous parking and charging.
- Continuous self-diagnosis and readiness checks, even when stored without signal.
- Local-first data handling: all data is processed and stored on the vehicle with end-to-end encrypted access via the app or onboard interface; camera covers and microphone off-switches are standard physical safeguards.
Safety, standards and the privacy promise
Tensor states the Robocar is purpose-built for SAE Level 4 autonomy and engineered to meet or exceed a wide range of global safety standards: FMVSS, UN/ECE, and GSO standards including MOIAT conformity. Tensor aims for top third-party ratings such as IIHS Top Safety Pick+, Euro NCAP and US NCAP 5-Star. To eliminate single points of failure, the vehicle’s design layers redundancy across:
- sensors, compute, communications, drive-by-wire, power, thermal management and control — enabling fail-operational performance;
- sensors and vision systems are protected and maintained by washer systems and intelligent covers to guarantee visibility in harsh climates.
Privacy is built in, not bolted on. Tensor’s stated principles:
- Your Robocar. Your data. Location, preferences and records are processed and stored locally on the vehicle.
- End-to-end encrypted access via the owner’s app or onboard interface.
- Physical privacy controls such as camera covers and microphone off-switches to reduce risk associated with always-connected vehicles.
About Tensor
Founded in Silicon Valley in 2016 and headquartered in San Jose, Tensor is a leading American AI company specializing in agentic products designed to empower individual users. With additional offices in Barcelona, Singapore, and Dubai, Tensor focuses on pioneering autonomous vehicle technology built from the ground up for private ownership at scale.Tensor’s ecosystem includes prominent partners such as Nvidia (providing an onboard supercomputer delivering 8,000 trillion operations per second), Sony (17-megapixel automotive-grade cameras), VinFast (advanced manufacturing), Oracle (cloud infrastructure), and Marsh, which supports the world’s first insurance policy tailored for Robocars.
Level 4 Autonomy, explained (the practical meaning)
SAE defines six levels of vehicle automation from Level 0 to Level 5. For context:
- Level 0 – No Automation: Human does all driving; basic aids (e.g., emergency braking) may intervene.
- Level 1 – Driver Assistance: One driving function automated, such as adaptive cruise or lane-keeping.
- Level 2 – Partial Automation: Steering and speed control can be handled together, but the human must stay engaged (examples: Tesla Autopilot, GM Super Cruise).
- Level 3 – Conditional Automation: The vehicle can handle many driving tasks but still requires human fallback readiness in case of system limits.
- Level 4 – High Automation: A vehicle performs all driving tasks within a defined operating design domain (geofence, urban area, dedicated lane) without requiring a human fallback in most conditions. This is where Tensor places the Robocar: designed to take full charge of driving tasks within its operating environment.
- Level 5 – Full Autonomy: The vehicle can operate anywhere and in any condition without human intervention — still a future target for mass production.
The difference between Level 3 and Level 4 is control and responsibility: Level 4 systems are built to operate autonomously in their domain without depending on a ready human operator as the fallback. Go to Source
