This post is also available in:
עברית (Hebrew)
As electric and autonomous vehicles take on more driving tasks, understanding exactly how a car behaves at any given moment becomes critical. One of the hardest variables to track is the sideslip angle—the degree to which a vehicle slides sideways during sharp maneuvers or on low-friction surfaces. This parameter plays a key role in stability control and accident avoidance, yet it cannot be measured directly with standard onboard sensors. Existing estimation methods rely on simplified physical models or indirect calculations, which often lose accuracy when road conditions, speed, or tire behavior change.
A new research effort proposes a more robust solution by combining physics-based modeling with artificial intelligence. The system uses what researchers describe as a “physical AI” approach: instead of replacing vehicle dynamics with a black-box model, it integrates real-world sensor data with established physical equations and machine learning. The result is a real-time estimation method that remains accurate across a wide range of driving conditions.
At the core of the technology is a hybrid estimation framework. A physical tire and vehicle motion model provides structure and interpretability, while an AI-based learning component compensates for nonlinear effects that traditional models struggle to capture. According to TechXplore, the researchers implemented this using an unscented Kalman filter to track vehicle states, combined with Gaussian process regression to learn complex tire behavior from data. This pairing allows the system to adapt quickly while maintaining the reliability expected from physics-based control systems.
Testing on an actual electric vehicle platform showed that the approach delivers high accuracy under varying road surfaces, speeds, and cornering scenarios. Faster and more precise estimation of sideslip angle improves how stability control systems respond, helping vehicles maintain traction and follow intended paths more safely. The method also supports more efficient energy use by enabling smoother, better-informed control decisions.
Beyond civilian mobility, the technology has clear defense and homeland security relevance. Military ground vehicles—especially electric or hybrid platforms—often operate on unpredictable terrain where traction loss can be dangerous. Autonomous or remotely operated vehicles used for logistics, patrol, or reconnaissance require reliable state awareness to function safely without constant human input. A vehicle control system that can accurately assess slip and stability in real time supports safer operation in off-road, damaged, or contested environments.
The researchers see the work as a foundation for future vehicle control systems rather than a single-use solution. By tightly coupling AI with physical models, the approach avoids many of the trust and robustness issues associated with purely data-driven systems. As autonomous driving expands into more complex scenarios, this hybrid method offers a path toward safer, more predictable vehicle behavior—whether on public roads or in demanding operational settings.
The research was published here.
