Researchers from the Norwegian University of Science and Technology (NTNU) are exploring how artificial intelligence (AI) can revolutionize driver training. In collaboration with IT firm Virtur and WAY AS, Norway’s only simulator-based driving school, they have developed an AI system that serves as a virtual driving instructor within a full-scale car simulator.
Neutral and Immediate Feedback
This AI instructor can evaluate learner drivers' performance instantly and provide unbiased feedback, minimizing any personal biases that human instructors might unintentionally introduce. Odd Erik Gundersen, an associate professor at NTNU, asserts that AI instructors can systematically assess learner drivers as effectively as human ones.
Addressing Road Safety Concerns
With road traffic accidents claiming nearly 1.2 million lives annually—especially affecting young drivers—improving driving instruction is crucial. The researchers emphasize that even minor enhancements in driver training could significantly save lives. Norway reported 118 traffic fatalities in 2023, prompting authorities to intensify traffic safety measures.
Comprehensive Training Scenarios
The AI system incorporates 477 different driving scenarios, focusing on three key traffic situations: overtaking, navigating urban traffic, and handling roundabouts. The goal is for the virtual instructor to accurately assess these scenarios and provide precise feedback, similar to what real instructors would offer.
In simulator sessions involving 21 learners and seven instructors, the AI's assessments closely matched those of human experts, demonstrating that it can operate with comparable precision.
Customizable Learning Environment
One of the advantages of simulator-based training is the ability to create various driving environments tailored to individual learners. The AI can modify scenarios based on the learner’s needs, helping them practice in safe, controlled conditions that might be hazardous in real life, such as driving in extreme weather or dealing with unpredictable obstacles.
Cost-Effective and Personalized Training
The researchers believe that transitioning to AI-driven instruction can reduce the costs of driving lessons. The system combines a rules-based framework with advanced AI algorithms to provide feedback and guidance that is easy to understand and act upon. This method is distinct from pure machine learning, which typically requires extensive data to train.
Technological Innovations
Using a moving platform surrounded by projectors, the simulator immerses learners in a realistic driving environment, capturing their actions and responses, including eye movements. NTNU plans to enhance the system with mixed reality (MR) technology, which merges virtual and augmented reality to create even more engaging training experiences.
Regulatory Challenges
While the potential for AI-driven driver training is immense, existing regulations in Norway mandate that instruction must occur in a real vehicle, limiting the use of simulators. To fully integrate AI instructors into the driver training framework, these regulations would need to be updated.
In summary, this research highlights the transformative potential of AI in driver education, paving the way for more effective, efficient, and safer training methods.