Luis Figueredo
Research Essentials for a Successful Deadline
Holistic Human-Robot Collaboration
My vision of robotics is an all-inclusive one. Particularly, when it comes to human-robot collaboration, a comprehensive view and understanding of both robot and human capabilities and limitations is critical for success.
A perfectly designed planner that comprehends and takes into account different human factors can become useless if the robot is unable to satisfy its own constraints and/or execute the individual subtasks and/or connections. At the same time, executing a collaborative task (or even a co-existence one) optimally (from the robot perspective) but without thinking about human safety, comfort, and preferences may lead to poor interaction, a lack of motivation from the human side, and even rejection to the robot counterpart — which defeats the purpose of HRC.
My research takes an integrative approach, considering both robots and humans. My approach is a bottom-up one, providing first real-time collision-free (when possible) planners and controllers that optimize for safety and performance, satisfying geometric and force constraints while ensuring a general understanding and satisfaction of tasks and constraints. This involves considering human safety at all levels of actions and the capability of using multiple arms to plan more complex, sequential, and parallel tasks. Finally, taking all these aspects into account, we can integrate human implicit and explicit communication, preferences, and comfort.
The topics below are critical for this view. The tabs will be updated whenever possible, and if not, feel free to check my previous presentations (see above) or just contact-me and have a chat.
Control Theory!
Control and modelling of dynamic systems is at the core of robotics research (and, many other applications).
With the advance of Lyapunov’s control and Kalman’s estimation theories, and the development of feedback systems and Wiener’s cybernetics systems, robotics flourished as a science of perceiving and manipulating the physical world with computer-based mechanical devices. More recently, control still plays a significant role in bringing machines closer to humans, ensuring compliance, safety and performance and constraint guarantees.
I am interested in many aspects of control theory, but particularly in understanding and advancement of control techniques for robot manipulators in the sense of reactiveness robustness, optimality and flexibility—cornerstone features for the required level of performance and robotic autonomy.
Example of Research Topics:
-
Robust control: (Automatica-2020), (ICRA-13)
-
Hybrid control: (JFI-2017), (IROS-2014)
-
Optimal control: (IROS-2015)
-
Time-delayed control (TAC-2017)
Example of videos about control theory
Geometric Methods and Dual Quaternion Algebra
In the study of body kinematics and dynamics, a suitable representation is fundamental for the effective description and solution of the control problems. For instance, several rigid motion representations detach and individually address attitude and translational dynamics. Nonetheless, detached approaches neglect the coupling between the full rigid motion dynamics—that is, the orientation and position coupling—which often yields an improper description of the rigid body motion. For rigid body transformation, I mostly use unit dual quaternions, group Spin(3) ⋉ R3, which has many advantages compared to other unified representations such as homogenous transf. matrices, HTM, (easy to describe, less computationally complex and expensive, among others). Furthermore, it is straightforward to extract geometric parameters (translation, axis of rotation, angle of rotation) and compute distances (line-to-line, to plane, to point etc). This is critical to define and constraint tasks (such as carrying a cup upwards, and keeping it away from the table, and self-collision). In my work, we explore such geometric relationship (named primitives) as well as corresponding distances and their mappings to robot kinematics and dynamics.
Robot Safety: Constraint Satisfaction and Robustness
Still under construction
Example of Research Topics:
-
Singularity Avoidance: (Automatica-2020), (ICRA-13)
-
Hierarchical control for multiple tasks via task-relaxation: (JFI-2017), (IROS-2014)
-
Quadratic programming methods (VFIs, CBFs, time2bound): (IROS-2015)
-
Multimanifold constraints (task, joint and manipulability): (TAC-2017)
