This research focused on preventing collisions between cyclists and heavy goods vehicles (HGVs). A collision avoidance system, designed to avoid side-to-side collisions between HGVs and cyclists, is proposed. The cyclist's motion relative to the HGV is measured with an array of ultrasonic sensors. The detected distances from ultrasonic sensors are processed in real time to construct a smooth trajectory for the cyclist. The controller assumes constant acceleration and constant yaw rate for both the HGV and the cyclist and extrapolates the relative motion forward in time. The HGVs' brakes are engaged if a collision is predicted. A prototype system was built and fitted onto a test truck. The proposed collision avoidance system was tested in real time and proved to be effective within certain speed ranges.

Heavy goods vehicles turning left across the path of a cyclist cause approximately 40% of the most serious injuries and fatalities to cyclists. A collision avoidance strategy is proposed in this study. Based on measurement of the current position and motion of the cyclist relative to the heavy goods vehicle, the future position of the cyclist after a short period is predicted by assuming constant acceleration and constant yaw rate. The brakes on the heavy goods vehicle are applied if the predicted future cyclist position overlaps the future envelope of the heavy goods vehicle. A parametric study shows that a high portion of potential collisions could be prevented in this way. Nineteen fatal accidents involving left-turning heavy goods vehicles and cyclists are reconstructed in simulation from police report files and the collision avoidance system is simulated. It is concluded that the introduction of the proposed collision avoidance strategy could possibly have prevented up to 18 of the 19 deaths in these accidents.

No abstract

This paper focuses on measuring the motion of a cyclist moving adjacent to a heavy goods vehicle (HGV), from the detections of ultrasonic sensors installed along the side of the vehicle. The measurements are used in a prototype collision avoidance system that predicts the future relative motion and assesses the likelihood of a collision. An array of ultrasonic sensors is adopted to cover the near side of the HGV, where most of fatal collisions with cyclists occur. A method combining quadratic programming and Kalman filtering is developed in this paper for recovering the bearing angles of the cyclist from the detected distances provided by off-the-shelf ultrasonic sensors. The algorithms are developed for use in real time and practical constraints are considered. The simulation and testing results prove the effectiveness of the proposed method for a reasonable range of the speed differential between the cyclist and the HGV.

No abstract