Robot design and construction

The group designs and constructs innovative mechatronic devices based mainly on parallel architectures. Our developments include the “Wrenchpad” (a six-axis tactile pad), several tensegrity-based robots, a pentaglide, several variations of the Gough-Stewart platform, different cable-driven robots, and the "Scherbot" robot (a five-bar mechanism to test kinodynamic motion planning and control techniques). The group also works on the development of various reconfigurable robots. These offer the possibility of reducing the number of actuators needed to perform a task, with the consequent decrease in construction costs. Moreover, reconfigurations can also be used to enlarge the robot’s workspace, or to avoid problematic configurations like singularities.

→ more info

Position analysis of multibody systems

The group also develops techniques for position analysis of multiloop linkages. The problem consists in finding the possible configurations that a linkage can adopt while respecting the kinematic constraints imposed by its joints. The resulting techniques can be applied to robotics (in contexts like direct or inverse kinematics, cooperative manipulation, object grasping, and motion planning), to structural biology (e.g., to the conformational analysis of biomolecules), to multibody dynamics (initial position and finite displacement problems), and to computer-aided design (variational CAD and assembly positioning). The group works essentially on two approaches: one based on relaxation techniques, and the other based on characteristic polynomials using Distance Geometry. Many of the developments are implemented in the CUIK suite, a large toolbox for motion analysis and synthesis of closed-chain multibody systems.

→ more info

Singularity analysis

Singularities play a prominent role on understanding the configuration space of a robot. Depending on their nature, singularities give rise to overspeeding problems, dexterity losses, or controllability issues. Thus, these configurations are often avoided during the usual operation of a robot, especially in applications that require careful human-robot interactions. Singularities, however, may also give rise to mechanical advantage (e.g., they can be used to transform small motor torques into large end-effector forces) and also provide the boundary of the workspace, which is a crucial information for the robot designer. The group has developed new geometric tools that allow characterizing and computing the various singularity loci of a robot, either for specific classes of parallel manipulators, or for general multi-body systems. New algorithms for controlling the motions across forward singularities are being developed too, which would allow the extension of the reachable workspace in parallel mechanisms.

→ more info

Motion planning and control

Along this line, the group develops algorithms for the planning and control of motions of general constrained systems. These systems encompass robots subject to holonomic or nonholonomic constraints, like loop-closure, contact, or rolling constraints. The goal is to design feasible motions bringing the robot to a desired state without colliding with obstacles, and to obtain robust controllers to perform such motions. Several techniques have been developed to both ends, which either consider the kinematic constraints of the robot, or also the full dynamic model (including motor saturations and speed limits). In both cases, innovative methods based on higher-dimensional continuation and randomised sampling techniques have been proposed for the planning of motions. The control of motions, in turn, is achieved by means of optimal control and trajectory optimization techniques. The group has also investigated the connections with related problems in biochemistry, contributing with novel algorithms for finding low-energy paths between different molecular conformations.

→ more info

Kinematic and Robot Design Laboratory

The Kinematics and Robot Design Laboratory was created thanks to the financial support of the VALTEC program, co-financed with FEDER funds, of the Autonomous Goverment of Catalonia. It was initially created to validate the practical interest of our parallel robot designs, but it has rapidly derived into an active lab where the prototypes designed by the researchers of the Group of Kinematics and Robot Design are implemented as proofs of concept.