Tesla is set to reveal its long-awaited "robotaxi" prototype on Thursday night, potentially marking a major step forward in its pursuit of self-driving vehicles.

This unveiling is a key moment for the company after years of unfulfilled promises by CEO Elon Musk to deliver self-driving vehicles. The automaker is expected to present a prototype called the "Cybercab," though it is likely not yet a fully operational, road-ready driverless taxi.

Tesla faces a considerable challenge in convincing regulators and the public of the safety of its self-driving technology. This task could take significantly longer than its competitors' progress, as firms like Alphabet's Waymo have already begun operating robotaxi fleets in select cities.

Unlike other major players in the industry, Tesla follows a distinct strategy that relies on "computer vision" and end-to-end machine learning, using cameras and artificial intelligence to make driving decisions. In contrast, other companies, such as Waymo, Amazon's Zoox, and General Motors' Cruise, combine this technology with additional layers of sensors like radar and lidar to improve safety and meet regulatory standards.

While Tesla's simplified approach is cheaper, it presents critical vulnerabilities. Experts and former Tesla engineers note that without additional layers of sensors and mapping used by its rivals, Tesla's system struggles with rare and unpredictable driving scenarios, known as "edge cases." These rare situations pose significant difficulties for autonomous systems, leaving Tesla's technology more prone to error.

Another challenge lies in Tesla's reliance on end-to-end AI, which operates as a "black box." One former Tesla engineer explained that this makes it difficult to pinpoint the cause of system failures, complicating efforts to safeguard against accidents.

Despite these obstacles, Tesla remains committed to its vision of fully autonomous robotaxis. Musk has indicated that the company is using the same Full Self-Driving approach in its robotaxi development efforts, though regulatory approval and overcoming technological hurdles may still be a long way off.

(With input from Reuters)