Year: 2020
Type: dissertation
Abstract: This thesis deals with the design and the implementation of a path planner for a four-wheeled, pseudo-omnidirectional robot. The path planning outlined in this work is based on the behaviour of a human driver faced with the same problem as the robot. Paths with low computational costs, suitable for a realisation in real-time, are being yielded and are being chosen over time-optimal paths. First of all a kinematic model has to be developed, which will be employed to a parking algorithm. \nThe parking algorithm is based on the behaviour of a human driver. For this purpose, the whole path is divided into two phases called A and B. The trajectory is calculated in reverse order. Starting from the parking configuration in phase B, a path with directional changes is computed until the vehicle can reach the starting configuration in one draw. This last draw is called phase A. In the end of the calculation the trajectory is reversed. \nBy applying the designed algorithm, the mobile robot approaches the target with the aid of sensor data. The target is approached bit by bit for every calculation step. Thus, the manner of approach is comprehensible intuitively for humans. \nFor every partial trajectory, the algorithm provides points in the plane, which are connected by B-splines of third order. The dynamic properties of arc length parametrised curves, such as velocity and acceleration, can be defined by specifying the profile of the arc length, which serves as the path parameter. This is achieved by a sin2-profile for the jerk of the trajectory. \nThe algorithm can be extended to calculate a trajectory for a towing vehicle with n trailers. For this purpose, it is necessary to develop an extended kinematic model, which can be employed for the extended algorithm. A system with a towing vehicle and only one trailer already leads to problems which need to be considered in the algorithm. \nThe finished path planner runs as a real-time process on a Linux-operating system with RTAI on a mobile robot of the Institute of Robotics of the Johannes Kepler University Linz.
Author: Rumenovski, Martin
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7023762147
