In the last decade, research studies on parking planning mainly focused on path planning rather than trajectory planning. The results of trajectory planning are more instructive for a practical parking process. Therefore, this paper proposes a trajectory planning method in which the optimal autonomous valet parking (AVP) trajectory is obtained by solving an optimal control problem. Additionally, a vehicle kinematics model is established with the consideration of dynamic obstacle avoidance and terminal constraints. Then the parking trajectory planning problem is modeled as an optimal control problem, while the parking time and driving distance are set as the cost function. The homotopic method is used for the expansion of obstacle boundaries, and the Gauss pseudospectral method (GPM) is utilized to discretize this optimal control problem into a nonlinear programming (NLP) problem. In order to solve this NLP problem, sequential quadratic programming is applied. Considering that the GPM is insensitive to the initial guess, an online calculation method of vertical parking trajectory is proposed. In this approach, the offline vertical parking trajectory, which is calculated and stored in advance, is taken as the initial guess of the online calculation. The selection of an appropriate initial guess is based on the actual starting position of parking. A small parking lot is selected as the verification scenario of the AVP. In the validation of the algorithm, the parking trajectory planning is divided into two phases, which are simulated and analyzed. Simulation results show that the proposed algorithm is efficient in solving a parking trajectory planning problem. The online calculation time of the vertical parking trajectory is less than 2 s, which meets the real-time requirement.

The traditional signalized intersection is no longer able to meet the increasing demands of transportation. A method combining autonomous intersection management (AIM) and connected and automated vehicles (CAVs) is suggested for solving existing traffic problems. This study proposes a control method for unsignalized intersections based on cooperative grouping, which discretizes the continuous traffic flow into a series of small-scale vehicle groups. In addition, a Petri Net model with a progressive structure is proposed, for which the time domain sequence of each group is strictly defined. To realize the global control of CAVs, the whole process of vehicle movement is decomposed into two processes: i ) adjustment on branches and ii ) coordination in the junction. All grouping conditions in the junction are clearly defined, and time windows are used for the discrete traffic flow modeling. Furthermore, a cooperative adaptive cruise control (CACC) model with bifurcations is proposed for grouping and following the control of vehicles on each branch. Simulation results show that vehicles are automatically grouped according to their initial positions, and no collision occurs during the whole process. Applied with the proposed method, the traffic capacity of a four-branch intersection exceeds 10000 veh/h under ideal conditions. The vehicles run more smoothly and show a better fuel economy than the other two methods.

In recent years, eco-driving strategies based on connected vehicle (CV) technologies have been studied to assist human drivers to reduce fuel consumption and pollutant emissions. In this paper, a real-time eco-driving strategy for CVs that considers human driver error is proposed to improve both traffic and fuel efficiency at signalized intersections where CVs and human-driven vehicles (HDVs) coexist. Firstly, a human driver error estimation model is established using real-world driving data. Then, based on the signal phase and timing information, vehicle state information, and the estimated human driver errors, a constrained nonlinear optimal control problem (OCP) is proposed to calculate the optimal advisory speed of each CV. The trajectory of HDV is estimated by utilizing the Gipps’ car-following model. Fast stochastic model predictive control (SMPC) is employed to solve the proposed OCP effectively. At last, simulation studies and real-vehicle experiments are conducted in various scenarios to verify the performance of the proposed strategy. Simulation and experiment results indicate that compared with the baseline strategies, the proposed eco-driving strategy can significantly reduce travel time and fuel consumption while ensuring the real-time performance.

This study presents an optimal control strategy for connected and automated vehicles (CAVs) at unsignalized intersections. A control framework with task-decomposition is designed for the near-intersection area. Firstly, the continuous traffic flow is discretized with a proposed intersection area-division scheme. Then, vehicles in different lanes are regrouped based on longitudinal positions to establish multivehicle cooperative optimal control problems of limited vehicle number. The expected optimal trajectory in the merging zone is obtained through offline simulation, which simplifies the constraints of the optimal control problem and improves the real-time performance of the online trajectory solving. Finally, an ideal capacity model is proposed based on the transit time window of vehicle, so that the capacity of a single lane is increased to 2400veh/h. The simulation results demonstrate that the proposed strategy ensures the driving safety at the unsignalized intersection, and this strategy efficiently reduces planning time for vehicle trajectories.