Iterative Road Grade Estimation for Heavy Duty Vehicle Control

Abstract: This thesis presents a new method for iterative road grade estimation based on sensors that are commonplace in modern heavy duty vehicles. Estimates from multiple passes of the same road segment are merged together to form a road grade map, that is improved each time the vehicle revisits an already traveled route. The estimation algorithm is discussed in detail together with its implementation and experimental evaluation on real vehicles. An increasing need for goods and passenger transportation drives continuing worldwide growth in road transportation while environmental concerns, traffic safety issues, and cost efficiency are becoming more important. Advancements in microelectronics open the possibility to address these issues through new advanced driver assistance systems. Applications such as predictive cruise control, automated gearbox control, predictive front lighting control and hybrid vehicle state-of-charge control benefit from preview road grade information. Using global navigation satellite systems an exact vehicle position can be obtained. This enables stored maps to be used as a source of preview road grade information. The task of creating such maps is addressed herein by the proposal of a method where the vehicle itself estimates the road grade each time it travels along a road and stores the information for later use. The presented road grade estimation method uses data from sensors that are standard equipment in heavy duty vehicles equipped with map-based advanced driver assistance systems. Measurements of the vehicle speed and the engine torque are combined with observations of the road altitude from a GPS receiver in a Kalman filter, to form a road grade estimate based on a system model. The noise covariance parameters of the filter are adjusted during gear shifts, braking and poor satellite coverage. The estimated error covariance of the road grade estimate is then used together with its absolute position to update a stored road grade map, which is based on all previous times the vehicle has passed the same location. Highway driving trials detailed in the thesis demonstrate that the proposed method is capable of accurately estimating the road grade based on few road traversals. The performance of the estimator under conditions such as braking, gear shifting, and loss of satellite coverage is presented. The experimental results indicate that road grade estimates from the proposed method are accurate enough to be used in predictive vehicle control systems to enhance safety, efficiency, and driver comfort of heavy duty vehicles.