Sensor Fusion for Predicting Vehicles' Path for Collision Avoidance Systems

Abstract

Path prediction is the only way that an active safety system can predict a driver's intention. In this paper, a modelbased description of the traffic environment is presented-both vehicles and infrastructure-in order to provide, in real time, sufficient information for an accurate prediction of the egovehicle's path. The proposed approach is a hierarchical-structured algorithm that fuses traffic environment data with car dynamics in order to accurately predict the trajectory of the ego-vehicle, allowing the active safety system to inform, warn the driver, or intervene when critical situations occur. The algorithms are tested with real data, under normal conditions, for collision warning (CW) and vision-enhancement applications. The results clearly show that this approach allows a dynamic situation and threat assessment and can enhance the capabilities of adaptive cruise control and CW functions by reducing the false alarm rate.

Bayesian environment representation, prediction, and criticality assessment for driver assistance systems

“…• Direct evaluation of the prediction equations of the tracking filter without measurement updates [20,22,163,200].…”

“…An obvious variant of extending the scope of the mentioned short-term trajectory models is to include knowledge about driving lanes or road boundaries as realized in [162,192,200]. These methods allow greater prediction times but do not permit the inclusion of additional assumptions about the actions of different traffic participants.…”

Collision avoidance on winding roads using dedicated short-range communication

“…Typical cooperative collision-warning systems were studied by Huang and Lin (2013); Lytrivis et al (2011); Polychronopoulos et al (2007), primarily based on a vehicle's dynamic state obtained from radar tracking, camera-based processing, or a communication device. In the advanced cooperative path-prediction algorithm proposed in (Lytrivis et al 2011), each vehicle can perceive its neighbours' positions, velocities, acceleration, headings, and yaw rate measurements through vehicle ad hoc networks.…”

“…In addition, drivers can be periodically informed of the present and future statuses of their neighbours. Similarly, (Polychronopoulos et al 2007) proposed a hierarchical-structured algorithm to fuse traffic environment data to accurately predict the trajectory of an ego-vehicle (vehicle equipped with sensors), allowing the active safety system to inform or warn the driver when critical situations occur. All abovementioned studies were based on updating the current state of a vehicle using motion models.…”

Semiautonomous Multivehicle Safety

“…Formal reasoning both for design and verification for autonomous vehicles driving in the presence of human drivers has been developed and implemented in the 2007 DARPA Urban Challenge by several of the participating teams [12]. Behavior prediction for human drivers has also been widely investigated (see, for example [24,31]). Yet, formally including these predictions into planning remains mostly an open question [12].…”