Behavior Recognition And Decision Scheme For Heavy-duty Vehicles On Roadways

The key points for designing unmanned commercial trucks and achieving intelligent auxiliary driving assistance system(ADAS)are to track and identify the driving behaviors of heavy-duty commercial trucks(HDV).The spatiotemporal information of trucks from trajectory and driving behavior can be employed to predict potential driving risks and make a safe driving,thus reducing road accidents.This paper presents an algorithm which can be used for trajectory tracking and behavioral recognition of HDV,and also can model the trucks’ movement on long inclined roadways.On this basis,this paper further focuses on the correlation analysis between driving behaviors and energy consumption from multiple dimensions,finding out the key factors affecting fuel depletion,to achieve the behavior decisions for energy saving,moreover,to enhance the interpretability of recommended decisions.To improve the accuracy on recognition due to the random noise in sampling data,this paper uses the time and semantics based trajectory segmentation method to obtain time-continuous samples at geographical locations.Specifically,Kalman filter is first used to distinguish error offsets from random noise and to estimate the elevation distribution of HDVs in different time intervals.With a real map,a Markov chain Monte Carlo model is then applied to classify truck behaviors based on the change in elevation between two geographical locations.As a result,the heavy commercial trucks movement model(HVMove)is proposed to recognize trucks behaviors according to the geometric structure of road network automatically.The HVMove model is verified and evaluated through extensive experiments based on a real-world trajectory dataset covering sections of an expressway and national and provincial highways.The results show us that HVMove model provides sufficient accuracy and efficiency for automated heavy-duty trucks and ADAS applications.At the same time,HVMove can generate maps with the elevation information marked automatically to information quantity and accuracy of traditional navigation map.Accurate behavior recognition is the basis of establishing decision system.This paper regards energy consumption of HDVs as the decision-making goal,constructs the behavioral framework from multiple dimensions,which aims to improve the interpretability and effectiveness of decision-making.This paper finds that there exists a huge number of multi-source original data cannot be calculated and represented as behavioral features.Therefore,this paper recognizes behavioral features of HDVs from the original data collected by sensors,and obtains the impacts of various features on fuel consumption through Bayesian classifier.Meanwhile,this paper uses the frequent mode to discover the correlation between different HDV behaviors to explain the behaviors,and optimize the driving decisions.What’s more,it analyzes the influence of speed and road on fuel consumption of HDVs,and discuss the correlations between driving behaviors and freight routes,which can ensure the ecological driving mode of HDVs with low fuel consumption and high safety.