01697385

Component

Trajectory optimization, a critical problem in connected autonomous vehicle control, has been intensively studied recently. A number of fast heuristic algorithms, such as shooting heuristics (SH) (1), have been developed to meet required time efficiency for real-time applications, but the optimality of their solutions is yet to be quantified. This paper aims to bridge this gap and compare the performance between fast heuristics and exact optimization models. For comparison purposes, the authors investigate a core trajectory optimization problem as a building block for a variety of trajectory optimization problems, i.e., guiding movements of connected automated vehicles on a general one-lane highway segment when the arrival and departure time and velocity of each vehicle are given. To enable the SH algorithm applicable to this general core problem, the authors adapt it to a fast-simplified shooting heuristic (FSSH) model that can solve the trajectory smoothing problems with different arrival and departure velocities. Then an exact trajectory optimization (ETO) model is formulated as a nonlinear programming problem, which takes vehicle position and velocity profiles as the decision variables, and the fuel consumption and the driving comfort of the whole platoon as the objective function. The constraints of the model are constructed based on vehicle dynamics limits and safety between consecutive vehicles. The authors prove the convexity of the ETO objective function, which ensures that the ETO model can be solved to the true optimum with the gradient descent algorithms supplied by the Matlab optimization toolbox. Further, 11 groups of numerical experiments with different input parameters are conducted. It is found that compared with FSSH, ETO can improve the objective values by a magnitude ranging from a few percent to tens of percent. However, FSSH far outperforms ETO in solution efficiency: the average solution time of FSSH is less than 1 second yet that of ETO is around 500 seconds.

This paper was sponsored by TRB committee AHB30 Standing Committee on Vehicle-Highway Automation.

19-04982

English

0000-0002-5264-3775

3p

2019

Transportation Research Board 98th Annual Meeting

Digital/other

References; Tables

Advanced vehicle control systems; Connected vehicles; Heuristic methods; Intelligent vehicles; Nonlinear programming; Optimization; Single lane traffic; Trajectory control

Data and Information Technology; Highways; Vehicles and Equipment

Transportation Research Board Annual Meeting 2019 Paper #19-04982

TRIS, TRB, ATRI

Dec 7 2018 9:26AM