Predicting Time Pressure of Powered Two-Wheeler Riders for Proactive Safety Interventions
This paper introduces MotoTimePressure (MTPS), a lightweight deep learning model that accurately predicts powered two-wheeler riders' time pressure using a new dataset of 129,209 feature windows, demonstrating that integrating these predictions significantly enhances collision risk forecasting and enables proactive safety interventions within intelligent transportation systems.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Every day, millions of people navigate the world on two wheels, from the bustling streets of India to quiet suburban roads. These riders face a unique set of dangers; unlike those inside cars, they have no metal cage to protect them, and their balance depends entirely on their own physical control. When a rider makes a mistake, the consequences are often severe. While factors like speeding and poor road conditions are well-known causes of accidents, there is a hidden force that quietly undermines safety: the feeling of being rushed. This sense of time pressure, the urgent need to reach a destination before a deadline, changes how a person thinks and moves. It narrows their focus, speeds up their decisions, and often leads them to take risks they would never consider when they are calm. For years, researchers have understood that this mental state is dangerous, but they have lacked a way to detect it in real time. Without a sensor that can "feel" a rider's stress, safety systems can only react after a mistake has already been made, rather than preventing it before it happens.
A team of researchers at the Indian Institute of Technology Indore has taken a significant step toward solving this problem by creating a system that can predict a rider's time pressure before it leads to a crash. To understand how this works, they first needed to gather data that simply does not exist in the real world, because asking people to ride dangerously on public roads to study stress is impossible. Instead, they built a high-fidelity motorcycle simulator, a stationary machine that looks and feels exactly like a real bike but is surrounded by giant screens showing a virtual city. They invited 51 experienced male riders, representing the vast majority of motorcycle fatalities in the region, to ride this simulator under three different conditions. In the first scenario, the riders had plenty of time to complete their journey. In the second, they were told to try not to be late. In the third, they were told the gate to their destination might close, creating a high-stress environment where they felt they had to rush.
As the riders navigated the virtual streets, the machine recorded every movement they made, from how hard they squeezed the brake lever to how they shifted gears. The researchers analyzed over 129,000 snapshots of this data to see how the riders changed when they felt time pressure. The results were stark. When the riders felt the most urgent pressure, their average speed jumped by nearly 50 percent compared to when they were calm. They braked much more violently, with front brake forces increasing by more than six times, and they made sudden, jerky stops far more often. They also took risky turns at intersections and failed to signal their intentions. Perhaps most telling was that even when the riders tried to compensate for their stress by being more careful, they still made more mistakes. The study found that under high pressure, riders committed 10 percent more dangerous errors and had 21 percent worse control over their vehicle, proving that human instinct alone is not enough to overcome the effects of a racing mind.
With this detailed picture of how stress alters behavior, the researchers built a new computer model designed to recognize these patterns. They named it MotoTimePressure. This model acts like a digital observer that watches the stream of data coming from the bike's sensors—speed, braking, steering, and gear changes—and instantly calculates whether the rider is calm, slightly stressed, or in a state of high pressure. The model was trained to spot the subtle signs of urgency, such as rapid throttle inputs or erratic steering, which often happen before a rider even realizes they are losing control. In tests, the system proved incredibly accurate, correctly identifying the rider's state more than 91 percent of the time. It is also remarkably efficient, small enough to run on a standard computer chip and fast enough to make a decision in less than a millisecond, which is fast enough to be used in real-time safety devices on a moving motorcycle.
The true power of this discovery lies in what happens after the system detects stress. The researchers showed that by feeding this prediction into other safety tools, they could dramatically improve the ability to forecast a crash. When they added the "time pressure" signal to existing collision prediction systems, the accuracy of those systems jumped significantly, getting almost as close to perfect prediction as if the system could magically read the rider's mind. This opens the door for a new kind of safety intervention. Instead of waiting for a rider to swerve or brake too hard, a smart motorcycle could detect the rising stress and gently warn the rider, perhaps with a subtle vibration or a calm voice prompt, suggesting they slow down or take a breath. This approach shifts safety from a reactive measure, which happens after an accident, to a proactive one that prevents the accident from occurring in the first place. By understanding and predicting the invisible weight of time pressure, this work offers a path to making two-wheeled travel safer for everyone, turning a fleeting feeling of urgency into a signal that can save a life.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.