Promoting excellence in mobility engineering

  1. FISITA Store
  2. Technical Papers

A Smart Method of Trajectory Planning for Automated Driving
FISITA2016/F2016-ACVD-005

Authors

Zhenfu Chen, Georg Roll

Continental, Division Chassis & Safety, Germany

Abstract

Research and/or Engineering Questions/Objective

For Automated Driving a reference trajectory is usually calculated by the autopilot to guide the vehicle. The motion control will attempt to force the vehicle into moving along this trajectory. The reference trajectory must be generated and continuously updated online during driving. Under considering the actual traffic circumstances, the reference trajectory should ensure safe, comfortable and normal vehicle behaviour. The questions here to answer are which properties should such a reference trajectory possess and how such a reference trajectory can be generated.

Methodology

To ensure normal vehicle behaviour the reference trajectory is laid out in the way that it consists only of three different essential elements which usually make up a normal road. These essential elements are straight lines, clothoids and circle arcs. By automated driving a temporary target status (target position or target ideal track) for the vehicle is normally defined according to the actual traffic circumstances. From the actual vehicle position towards the target status a reference trajectory will be so generated that it is smooth, continuous and differentiable according to the distance. To do this the trajectory generation is disassembled to three basic tasks, to correct the lateral offset, to correct the course angle difference and to correct the curvature difference between the actual vehicle position and the target status. Each basic task has to be completed by using only essential elements.

Results

In this way a reference trajectory can always be laid out which connects any actual vehicle position or status to a desired target status. Such a reference trajectory guarantees a human driver similar behaviour for the autopilot and vehicle system because it consists only of straight lines, clothoids and circle arcs just like the common traffic road. The generation of such a trajectory is simple and involves few calculations, so it is quite suitable for real time trajectory planning. The trajectory planning can optimise the active safety and the comfort by considering the vehicle speed and traffic circumstances. The maximal safe lateral acceleration will be kept by limiting the maximal curvature. The features of the steering actor will also be taken in count by limiting the graduate of the curvature.

Limitations of this study

The new method to generate trajectory can be used for nearly all possible driving situations, for straight ahead, curve driving, lane change or parking, even for reverse driving. The trajectory calculation procedure presupposes that certain target states (positions) are calculated by a suited road model and situation observer. If it turns out that the calculated trajectory cannot be followed by the vehicle due to exceeding the maximum lateral acceleration possible, a deceleration request will be generated to keep the vehicle within the physical limits.

What does the paper offer that is new in the field in comparison to other works of the authors

The use of only three essential elements to make trajectory planning is absolutely new and makes the trajectory generation easer, quicker and more precise in comparison to other methods. The main goal is to find the optimal combination of the essential elements to eliminate the possible the lateral offset, course angle difference and curvature difference between the actual vehicle status and the target status.

Conclusions

A smart method for trajectory planning is developed to generate the desired trajectory online by automated driving. The trajectory connects the actual vehicle position to any desired target status by using only three essential elements: straight lines, clothoids and circle arcs. Such a trajectory can be generated quickly and precisely .It is suitable for all possible driving manoeuvres. Moreover, this method enables safe, comfortable and driver-similar vehicle behaviour, if the vehicle exactly follows the generated trajectory.

Key Words: Trajectory; Automated Driving; essential elements; online generation; lateral offset; course angle; curvature

Add to basket