AGV Path Tracking
Existing algorithms for path tracking treat an AGV as a rigid point, describing its path as either a series of discrete points or a continuous curve. The problem with this approach is that it ignores factors such as the shape of the AGV or width of the road, possibly resulting in collisions or the AGV straying from the road.
As the researchers explain in their paper, AGV path tracking involves two problems: steering control and velocity control. To address the concerns of AGV shape and road width, the researchers were primarily focused on the steering control problem of path tracking, taking the vehicle’s velocity to be a constant.
The new steering control method considers the shape of the vehicle to be a rectangle, as opposed to a single point, and takes into account the width of the road as a boundary constraint. The researchers combined this model with the method of model predictive control (MPC) in an attempt to solve the path tracking problem.
The team tested their new approach using a Hongqi HQ430, an AGV with both an environmental perception system and driving control system. According to the researchers, the experiments demonstrated the effectiveness of applying their new method of path tracking to the HQ430, as the vehicle managed to follow a square path.
While this new method may provide one piece of the self-driving puzzle, improving our technology isn’t the only step to widespread adoption of driverless vehicles. For an overview of some of the political and philosophical barriers, read The Legislation, Liabilities and Ethics of Self-Driving Cars.