Brian Moore
Research Scientist
Dynamic balancing of linkages
Linkages which are dynamically balanced do not apply forces and torques on their base. This reduces undesired effect like vibrations. See the excellent description on the web page of the robotic lab of Laval University. For simple linkages, I am particularly interested in finding sufficient and necessary conditions on the design parameters of the bars of the linkage (length, mass, center of mass position, inertia) such that it is dynamically balanced, without adding any additional mechanical elements to the linkage. Up to now, our main results was a complete classification of dynamically balanced planar four-bar linkages and a proof that spherical four-bar linkages cannot be dynamically balanced without adding additional elements. A list of research papers on this topic can be found here.
Kinematics and Algebraic Geometry
Many problems related to robot kinematics can be formulated algebraically (i.e. as a system of polynomial equations). Using methods from algebraic geometry combined with computer algebra systems (i.e. symbolic computation tools), these systems of equations can be efficiently "solve". The meaning of solve depends on the problem. Typical challenging problems are: inverse kinematics of serial robots, forward kinematics of parallel robots, synthesis of mechanisms, analysis of linkages, ...
Biologically Inspired Robot Grasping
Although humans are very skilled at manipulation and in spite the recent developments in robotic manipulation, developing a robotic system that matches human hand dexterity is still elusive. One major impediment is the complexity of controlling such complex systems, due mainly to the large number of DOF. One way to overcome this problem is to take advantage of the biological similarities between the human and the robot. A new paradigm for obtaining skilled robot behavior is to utilize an intuitive teleoperation system where humans learn to control the robot joints to perform a given task. Once the human manages to control the robot and complete the desired task with it, the control commands produced by the human-robot system during task execution can be used for designing controllers that operate autonomously. This places the initial burden of learning on the human instructor, but allows the robot to ultimately acquire the ability to perform the target skill without human guidance. Using this paradigm, we are currently studying the reaching and grasping motion inferred by the human on the robot and how to generalize the motion to obtain a generic grasping controller. You can see videos of the virtual grasping after the training phase and of the corresponding motion on the robot (object 1, object 2)
