Vehicle parameter estimation using road surface vibrations

Sammanfattning: Traffic safety is a big concern of many modern societies. Every year, car accidents cause many injuries and fatalities. This has aected many car manufacturers and governments equally. While governments try to reduce the number of accidents by educating drivers or imposing regulations, car manufacturers have successfully incorporated diverse safety functions such as seat belts, anti-lock breaking systems, or airbags in their vehicles. Recent advances in communication technologies have given rise to new approaches for advancing vehicular safety even more: having vehicles and the road infrastructure communicating with each other will enable new safety systems that can also take the behavior of other road users into consideration. The information provided by the infrastructure stems from roadside sensors that continuously measure traffic and track vehicles. Parameters of interest are, among others, vehicle class or vehicle speed. Clearly, many sensors for estimating these parameters exist. However, these are often too limited, too expensive for maintenance, or not developed well enough in order to be deployed in large scale. For example, vision-based systems can provide very comprehensive information, as long as the line of sight is not obstructed and enough computational resources are available. On the other hand, miniaturized sensors are becoming more and more popular in conjunction with wireless sensor networks. This approach is also put forth in this work. The aim of the thesis is to examine the potential of using accelerometers mounted on the road surface for estimating parameters of vehicles passing the sensor. In the four research papers composing this thesis, it is shown that this novel approach is viable. First, the feasibility is analyzed based on measurements of real traffic and research challenges are identied. The two rst applications are derived from that: vehicle detection and wheelbase estimation, where the latter can only be achieved under the knowledge of a vehicle's speed. Then, the underlying mechanisms, namely wave propagation in the road is examined in a system identication framework. It is found that Lamb waves, that is, waves in a thin plate, are predominant and a model for describing this is proposed. Finally, an Extended Kalman Filter for vehicle tracking based on a moving constant force and the wave propagation model is proposed. As a result, estimations of the vehicle velocity and wheelbase are automatically obtained. It is indicated that future research should further rene the physical model of the source (vehicle) as well as the wave propagation and also improve the proposed Kalman Filter. Furthermore, there is still a lot of potential in exploiting other features of the measured signal.