Ego-Motion and Road Attitude Estimation Using Ground Speed Radars and an IMU

Abstract

Abstract Accurate ego-motion estimation is essential for autonomous and assisted driving in heavy-duty vehicles. This task becomes particularly challenging in scenarios where traditional sensors, such as wheel speed encoders, are unreliable. Additionally, a good understanding of the road characteristics, such as slope, is critical to ensure safe and efficient vehicle operation. This thesis addresses the mentioned challenges through a method that fuses measurements from ground-speed radars and an Inertial Measurement Unit (IMU). The proposed method jointly estimates vehicle motion and road attitude, while accounting for issues such as IMU bias and multirate sensor data. The main contributions of this thesis are the estimation of road slope angle and a method for estimating IMU accelerometer biases. The road slope is estimated through pseudo-measurements derived from the vehicle’s translational velocity. Meanwhile, the IMU accelerometer biases are estimated by employing a tilted IMU configuration. This setup enhances the observability of the biases, leading to improved filter convergence and estimation accuracy during typical driving maneuvers. The proposed method is evaluated using simulated data generated with IPG Truck- Maker (TM), a high-fidelity simulation environment that enables controlled, repeatable experiments across diverse driving scenarios. A Monte Carlo (MC) analysis is performed using multiple sensor configurations, allowing statistical assessment of system consistency and robustness. The results demonstrate that the filter reliably estimates vehicle velocity, attitude, and the road slope angle despite the presence of sensor noise and bias. These findings support the effectiveness of using a minimal radar setup in combination with a tilted IMU to enable robust and accurate egomotion estimation in heavy-duty vehicle applications.