| Summary: | This thesis presents a new trajectory planning algorithm for planning safe and smooth tra
jectories of a robot tool moving along a specified curve subject to application constraints.
The application constraints include geometric constraints (allowable orientation and po
sition of the tool) and motion constraints (joint limits, robot configuration inversions and
collision avoidance). The algorithm, which is based on a configuration space approach for
robot motion planning, addresses the problem of motion planning along a curve rather
than point-to-point motion associated with pick and place operation. Configuration
spaces are searched to assess the locations of the obstacles in the vicinity of the curve
and the curvature of the curve in the work space. The trajectory is planned by searching
between path segments connecting successive curve points, generating the path along
a curve linking the path segments, and selecting the optimal configurations along the
path to construct the final trajectory. If more than one paths exist, an A* search is em
ployed to optimize the path. The selected path specifies a sequence of tool configuration
ranges, and each configuration produces several inverse kinematic solutions. A second
A* search is used to select inverse kinematic solutions, which avoid major changes of
robot configuration (i.e. joint inversion). Since the algorithm searches the allowable tool
orientation range, the optimal trajectory which maximizes process quality is obtained.
The trajectory allows the tool to translate and rotate simultaneously without collision,
and the tool remains within the allowable orientation range and tracks the desired curve
within a given tolerance. Since the step size between curve points is varied dynamically
as the search proceeds, the path is planned based on a minimum number of curve points,
resulting in fast path planning. A smaller step size is used if the curvature of the curve
or the likelihood of collision is high. The primary advantage of the algorithm is not only
in finding a collision free path but in planning a trajectory which satisfies application
constraints. It is felt that this algorithm provides a high quality process which is difficult
to achieve with manual robot programming.
An off-line simulation program, TRAJPLAN, has been developed based on this al
gorithm. After the user inputs the application specifications, selects the curves to be
processed and the environment objects of interest, TRAJPLAN automatically plans the
trajectory of robot tool and generates the required set of sequential robot joint angles.
TRAJPLAN plans the motion with different precision depending on the requirements of
the application.
This algorithm has been demonstrated for robotic fish butchering and welding, and
the issues related to these applications are discussed in this thesis. As an example of a
typical problem, the system automatically plans the trajectory of a robot tool moving
along a 3-D curve with 31 points and surrounded by 5 obstacles in 1’20”. The algorithm
is suitable for a variety of robot applications which require the robot tool to move along
a curve, keep a certain orientation and position relative to the curve, and avoid collision
with the environment. === Applied Science, Faculty of === Mechanical Engineering, Department of === Graduate
|
|---|
