The long-term goal
is an autonomous warehouse with several surveillance cameras, robots and people
(e.g. for maintenance). The surveillance cameras support
the robots with navigation, including path planning. For the purpose of simplification, only the robot is considered in this work. The required paths are assumed to be known.
The aim of this thesis is to develop
a predictive, adaptive
path-following control. The predictive feature
should be
used to adapt the linear and rotational speed to the future situation (e.g. braking before a sharp bend). It is important to consider whether a constant or an adaptive (e.g. speed dependent) predictive
horizon should be used. In simplest case the predictive horizon is only
one look ahead point. For path following the path will be represent as a series of waypoints. Therefore, as pai1 of the control system, a decision must be made regarding when the next waypoints is adopted as the setpoints and when the end point
is reached. The evaluation be based
on travel time, distance, jerk, computing power and pose/position
error for several predefined paths.
A number
of previous works, functions, software architectures and a basic simulator will be
provided (including collision detection, speed controller,
etc.). Everything should be implemented in MATLAB/SIMULINK with a modular software design using reference
subsystems.
–
Familiarisation with the existing simulator and/or hardware and MATLAB/SIMULINK
– Familiarisation with path-following control, predictive path-following control, adaptive
control, the relevant
previous work
–
Literature research with subsequent evaluation of the methods
and extensions based on the given
criteria
– Development of a predictive, adaptive path-following control
system
– Implementation of the path following control
– Evaluation of the path sequence
control based on predefined paths
– Documentation of the work results that fulfils the requirements of the final thesis
