Technology reporter experiences the Mercedes-Benz CLA self-driving car equipped with the NVIDIA Hyperion platform.

TechRadar reporter Mike Moore personally experienced a Mercedes CLA automated driving sedan equipped with NVIDIA’s Hyperion platform during the GTC 2026 event. He said this self-driving journey was a successful experience—AI took full control with ease, could respond to the surrounding environment, and throughout the trip he felt safe. By integrating sensing hardware with end-to-end technology, Alpamayo has shown initial results in real-time decision-making and environmental perception on complex urban roads.

How do hardware sensors and end-to-end technology build a safety perimeter?

Mike Moore’s test drive experience, conducted in downtown San Jose, was in the Mercedes-Benz CLA model. The vehicle’s hardware specifications are built on the Hyperion 8 platform, with a total of 10 cameras and 5 radars positioned at both the front and rear of the vehicle, forming a 360-degree environmental perception network.

The software architecture is based on Alpamayo (End-to-End) end-to-end technology. A key feature of this technology is that it uses real-world road data and synthetic data to train models, with fully traceable technical records, aiming to improve the system’s safety and reliability. Although this complete system currently being experienced has not yet entered mass production, it is expected to be officially launched in the second half of 2026, when it will inject new technological momentum into the automotive market.

What are the interaction mechanisms of “Level 2 automated driving” on real roads?

The Hyperion 8 technology in the test vehicle belongs to Level 2 Autonomy “L2” automated driving. Under this classification, the system can take over navigation and steering, but the driver must still monitor road conditions. During the roughly 45-minute test drive, the car moved through urban and suburban environments with single-lane and multi-lane roads. The system design requires the driver to periodically touch the steering wheel to confirm they are alert and not distracted.

The driver has the lead at all times and can instantly disengage system control at any moment through actions such as pressing the brake pedal. In today’s regulatory framework, this human-machine cooperative mode is the main transition form for automated driving technology to enter the mainstream market—providing convenience while also ensuring that in sudden situations, the driver still has room to take control.

Mike Moore said that anyone who has ridden in an automated driving car knows that the experience may be stressful, especially when he was sitting in the passenger seat at the time. Americans call the passenger seat the Dead Seat, and reportedly the passenger seat is more prone to casualties than the driver’s seat; but after a few intersections, he relaxed and fully enjoyed the road-trip touring experience.

How does Alpamayo handle sudden road conditions and non-typical traffic behaviors?

During the test drive, Alpamayo demonstrated its capability in complex decision-making. When nearby-lane buses suddenly changed lanes to avoid an obstacle, the test vehicle immediately turned on its turn signal and moved synchronously to the adjacent lane to avoid a collision. In addition, the system also detects non-typical behaviors—for example, when it detects pedestrians preparing to cross a residential area street outside a designated crosswalk, the vehicle will slow down in advance and pull over to the side.

For route planning, the vehicle can enter the correct turn lane one block ahead to avoid dangerous behaviors at busy intersections, such as risky lane cutting or crossing lanes. This kind of predictive logic shows that artificial intelligence, when handling dynamic traffic flows, has already integrated map information with real-time sensing data to make judgments that better match traffic patterns.

The most challenging scenario during the testing involved encountering a large truck making an abnormal reverse maneuver in a parking lot and crossing the test vehicle’s lane. In this extreme case, the automated driving system demonstrated braking actions more timely than a human response, effectively avoiding a collision.

This article, “Tech reporter’s experience with the Mercedes CLA self-driving car equipped with the NVIDIA Hyperion platform,” first appeared in Chain News ABMedia.