Whole-body control of humanoid robots at second order kinematics under unilateral constraints

• Autor(a) principal: Juan José Quiroz Omaña
• Data de Publicação: 2021
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Repositório Institucional da UFMG
• Texto Completo: http://hdl.handle.net/1843/38700
• topic: Vector fields inequalities; Humanoid robots; Gauss's principle of least constraint; Dual quaternions; Quadratic programming; Engenharia elétrica; Desigualdades diferenciais; Processos gaussianos; Robôs – Sistemas de controle

This work uses vector field inequalities (VFIs) to prevent robot self-collisions and collisions with the workspace. We extend the VFIs to second order kinematics (SOVFIs) and, differently from previous approaches, the method is suitable for both velocity and torque-actuated robots. Furthermore, we present a formal proof of collision avoidance using SOVFIs. We propose a new distance function and its corresponding Jacobian in order to generate a VFIs to limit the angle between two Plücker lines. This new VFI is used to prevent both undesired end-effector orientations and violation of joints limits. In addition, we propose a new Jacobian related with the support polygon of a humanoid robot. This is used to maximize the support polygon area of the robot, and potentially increasing the robot's reachability and the robot safety in terms of its balance. We use the proposed Jacobians and the VFIs framework to enable whole-body control with multi-contacts using a full humanoid robot in simulation. The Euler-Lagrange model, which is used in conjunction with the SOVFIs for torque-actuated robots, is derived by means of the Gauss's Principle of Least Constraint using dual quaternion algebra. The use of dual quaternion algebra allows a more compact and unified representation for the twists and wrenches. The proposed method is evaluated in a realistic simulation on a 27-DOF full humanoid robot and on three real platforms: a 9-DOF humanoid robot, a 8-DOF bimanual manipulator, and a 16-DOF nonholonomic bimanual manipulator. Results show that all constraints are respected while the robot performs a manipulation task.