Trajectory planning

Print

The goal of this research program is to develop algorithms for trajectory planning of complex robotic systems.

Obstacle Avoidance

The first objective was to develop algorithms for the trajectory planning of redundant manipulators in an cluttered environment. A series of algorithms were thus developed, based on the discrete potential field approach applied in the Cartesian space. These algorithms lead to a simple and relatively robust trajectory planning. They were developed as part of a contract with the Hydro-Quebec Research Institute (IREQ). Figure 1 shows a simulator used for the development and verification of the algorithms. The operator can control the movement of the effector by specifying either a final destination only or specifying the local cartersian movement of the effector. In both cases, obstacle aviodance is automatically achieved by the algorithm.

Fig. 1: Simulator of the obstacle avoidance algorithms (for an enlargement, click on the image).

Simulator for the maintenance of power lines

The algorithms mentioned above were developed for applications involving the maintenance of live power lines by telerobotic systems. A complete graphic simulator was also developed for this application, so as to simulate the 23 degree of freedom system developed by IREQ. The IREQ system involves a 4 degree of freedom mast, two 7 degree of freedom SARCOS manipulators each monted on sliders and an auxiliary 3 degree of freedom arm. The complete system and working environment are shown in figure 2.

Fig. 2: Graphic simulator of the 23 DOF IREQ system (click on the image to view an enlargement).

Redundant robot trajectory planning

Another aspect involved the development of an algorithm for the trajectory planning of a redundant robot when certain joints are blocked. The algorithm then proceeds to redefine the kinematics of the robot to allow the application of conventional trajectory planning algorithms. This algorithm was optimized for the 7 degree of freedom manipulator SARCOS and integrated into the KALI controler used at IREQ.

Grasping of moving objects

The final aspect involved the development of approach and grasping algorithms which would allow a manipulator to grasp a moving object. A complete simulation of a system consisting of a manipulator and three cameras was developed, which allows a manipulator to track the trajectory of an object arriving into the workspace and to plan the movement of intercepting and grasping this object. This was achieved using the concept of generalized inverse kinematics.

Fig. 3: Simulator for the intercepting and grasping of objects. The field of view of each of the three cameras is represented by semi-transparent zones. The two fixed cameras are 2D cameras whereas the camera carried by the robot is a 3D camera. Click on the image to view an enlargement.

Applications

Trajectory planning of redundant rmanipulators for obstacle avoidance purposes has applications in all areas of robotics, such as operations involving assembly, maintenance, sodering, etc. Moreover, applications in teleoperation are numerous and cover many fields where operators are required to carry out dangerous tasks.

Approach and grasping actions are particularly useful for manipulators operating in non-structured environments, for example parts which are moving in an assembly line or on a conveyor and which must be grasped by a manipulator. Another example involves tasks where the objects to be grasped move due to external forces, such as the wind.