The model simplifies highway merging to improve understanding and advance autonomous vehicle technology.
Researchers at TU Delft have created a new model that provides a more accurate description of human behavior during motorway merging. Existing models typically assume that drivers continuously strive to optimize their actions to arrive at their destinations quickly and safely, an assumption that does not always hold true, according to postdoctoral researcher Olger Siebinga. This new model offers greater insights into how humans interact on motorways and has potential applications in enhancing autonomous vehicle technology.
For most drivers, merging onto a highway is a habitual task that doesn’t seem to require much consideration of the various factors at play. However, the complexity of this action becomes apparent only when attempting to replicate it within a computer simulation.
Simplified merging scenario
Siebinga, in collaboration with the team, introduces a pioneering interaction model rooted in risk perception and communication. This model is unique in its comprehensive explanation of human interaction dynamics at various levels, ranging from basic control inputs like acceleration, to the safety distances maintained between vehicles, and extending to decision-making processes about right-of-way. Such depth makes the model exceptionally relevant for applications in autonomous vehicle technology.
The genesis of this model traces back to a prior experiment led by Siebinga, involving a controlled merging task in a simulated environment. In this experiment, two participants could only accelerate or brake while separated by a barrier, relying solely on the visual information from a computer screen to guide their actions.
Understanding human behavior
Modeling enhances our grasp of human behavior. This is especially crucial in automated driving, where a major challenge is ensuring that human drivers comprehend and trust autonomous vehicles. Siebinga’s latest model sets the foundation for the safe and widespread acceptance of self-driving cars. He is now working to expand this model to incorporate steering dynamics.