Sensor-fusion reference governor control system for tractor-trailer rollover prevention

Background

The increasing deployment of heavy-duty tractor-trailer combinations in logistics and transportation has elevated the importance of advanced vehicle stability systems. Sudden maneuvers, uneven road surfaces, and varying payload distributions can generate high lateral accelerations that threaten rollover stability. With stricter safety regulations and growing reliance on advanced driver-assistance systems (ADAS), there is a pressing need for robust rollover prevention technologies that can proactively manage lateral forces and ensure safe cornering without extensive hardware retrofits.

Existing rollover mitigation approaches predominantly rely on trailer-mounted modules equipped with inertial measurement units and brake actuators. These add-on systems are costly to install and maintain, especially for fleets that frequently interchange trailers. Furthermore, they typically react only after a predefined acceleration threshold is exceeded, leading to delayed intervention. Many current methods also depend on detailed physical models that struggle to capture real-world uncertainties—such as variable loading or wind disturbances—and often overlook the amplification of lateral forces from tractor to trailer, resulting in constrained performance and limited reliability.

Technology

A tractor-mounted suite of sensors—including an inertial measurement unit, wheel-speed sensors and GPS—feeds a sensor-fusion algorithm that continuously estimates vehicle states such as lateral acceleration. A data-driven ultra-local model treats the combined tractor–trailer dynamics as a lumped-parameter system, with an estimated disturbance term and gain identified in real time via a Kalman filter. A feedback control law then predicts future lateral accelerations and passes a desired lateral-acceleration reference from an ADAS module into a chance-constrained reference governor. That governor adjusts the lateral-acceleration setpoint to account for rearward amplification of lateral forces and stochastic disturbances, ensuring the resulting tractor–trailer acceleration remains below a rollover threshold with a predefined probability. The modified reference drives a low-level steering controller, and actuator feedback is used to update the model and enforce constraints continuously.

This approach stands apart by eliminating trailer-mounted modules and relying solely on tractor-based sensors, reducing cost and simplifying fleet operations. The ultra-local model requires no detailed vehicle parameters, adapting in real time to changing conditions. The chance-constrained reference governor proactively modifies ADAS commands rather than reacting after threshold exceedance, and its probabilistic framework accounts for sensor noise, wind disturbances and mass variations. Together, these elements deliver a robust, scalable rollover-prevention system that maintains safety margins under real-world uncertainties.

Benefits

• Eliminates costly trailer-mounted modules by using tractor-based sensors, reducing fleet expenses and complexity
• Proactively prevents rollovers by adjusting steering commands before lateral‐acceleration limits are reached
• Adapts in real time via a data-driven ultra-local model and Kalman filter for accurate dynamics estimation
• Ensures lateral acceleration stays below the rollover threshold with a predefined probability using a chance-constrained governor
• Integrates seamlessly with existing IMU, GPS, and wheel-speed sensors—no additional trailer hardware required
• Enhances robustness to disturbances (sensor noise, wind, load variation) and works across varied tractor-trailer configurations

Applications

• Tractor-trailer rollover prevention
• Heavy-duty vehicle stability control
• Fleet safety management
• ADAS retrofit kit
• Insurance risk mitigation

Additional info

A tractor-mounted IMU, wheel-speed sensors and GPS feed a sensor-fusion algorithm to estimate states. A data-driven ultra-local model (lumped disturbance F and gain β estimated via Kalman filter) predicts lateral dynamics. A chance-constrained reference governor adjusts ADAS-provided lateral-acceleration commands, accounting for rearward amplification and stochastic disturbances, to ensure the tractor-trailer lateral acceleration stays below rollover thresholds with specified probability. The modified reference drives a low-level steering controller with feedback for model updating and constraint enforcement.