Unified Path Planning and Control of Autonomous Vehicles Using Artificial Potential Fields


Staff Mentor:

prof. J. Hellendoorn


Other Mentor(s):

Dr. Mohsen Alirezaei, Dr. Elham Semsar Kazerooni

Keywords:

Road traffic networks; Identification and estimation; Model predictive control

Description:

Introduction
The availability of driver assistance systems and autonomous functions has increased rapidly for production vehicles in recent years. These systems are being developed for active safety systems and aim at assisting the driver in all driving situations. To further develop the technology and get closer towards fully autonomous vehicles, several research projects have been initiated in TNO. Using physics-inspired methods like artificial potential fields (APF) for navigation control of autonomous vehicles is of interest. APF which has been extensively used in path planning and control of autonomous mobile robots, seems to be quite powerful framework to implement in autonomous vehicles.

Problem background
In most autonomous vehicles, navigation control is based on first planning a specific path to a goal and then tracking it using a feedback and feedforward controller. The potential function is an intuitive way to represent the level of hazard experienced by the vehicle. It allows to define a driving corridor with some acceptable tolerance as opposed to following a stringent trajectory. Furthermore, APF can be used to integrate path planning and motion control in navigation control of autonomous vehicles. Design of a unified path planning and control of autonomous vehicles using the APF method is the main contribution of this assignment.

Your assignment
Literature review
Design of a motion planner and controller based on APF method by:
- Modelling the occupancy map of the road and its characteristics, such as lanes, using
potential functions.
- Formally defining all requirements on safety, comfort and performance of the autonomous
vehicles.
- Find an appropriate controller which minimizes the potential function subject to the given
requirements.
Evaluation of the APF in simulation considering the real time implementation constraints
Verification of the APF in experimental tests in case of promising simulation results
Results are to be presented in a PowerPoint presentation and reported in a report.



Original description

© Copyright Delft Center for Systems and Control, Delft University of Technology, 2017.