Drive Force and Longitudinal Dynamics Estimation in Heavy-Duty Vehicles.

Area of Science:

• Engineering
• Computer Science
• Transportation Systems

Background:

• Dynamic modeling of heavy vehicles is crucial for applications like autonomous driving and simulation.
• Acquiring and integrating data from various vehicle subsystems presents challenges due to differing signal rates and potential data loss.

Purpose of the Study:

• To develop a non-invasive data acquisition system for heavy vehicles.
• To implement a data fusion technique for combining multi-rate and incomplete sensor data.
• To enable accurate vehicle identification and dynamic modeling.

Main Methods:

• A non-invasive hardware/software setup was used to collect data from an urban bus.
• Data from the vehicle's CAN bus, IMU, GPS, and pedal sensors were gathered wirelessly.
• A Kalman filter was employed for non-conventional sampling data fusion.

Main Results:

• The proposed system successfully collected data from multiple internal vehicle networks and sensors.
• The Kalman filter effectively fused data from disparate sources, handling different sampling rates and missing data points.
• Accurate signal estimation was achieved, facilitating vehicle identification and modeling.

Conclusions:

• The combination of novel data acquisition and multi-rate filtering is effective for heavy vehicle dynamic modeling.
• This approach provides a robust solution for utilizing sensor data, even under challenging conditions like data loss.
• The methodology supports advancements in driving simulation, autonomous systems, and vehicle analysis.