Title: Direct force control in ROS2

Author of the experiment: Giovanni Intonti

Description: Force control is fundamental in industrial tasks requiring accurately exerting a desired force profile. To this aim, this thesis proposes a ROS2 robot-agnostic direct force controller (DFC). The developed controller manages parallel motion and force references. The former are managed by a Cartesian pose controller (CPC) exploiting a closed-loop inverse kinematics resolution, while the latter are tracked by a force feedback loop, outputting motion references to the chained CPC. The controller is validated in both simulated and real scenarios, including interacting with horizontal and vertical environments, and a teaching-by-showing human-robot collaboration experiment.


Title: Admittance control in ROS2

Author of the experiment: Francesco D'Onofrio

Description: A fundamental challenge in robotics is ensuring that robots can safely and effectively interact with their environments. To empower robots with this capability, an admittance controller is developed in ROS2. Compared to the ros2_control native one, such controller allows to manage task space references, which are rather necessary when online redundancy resolution is needed, and supports the execution of offline-planned trajectories. Furthermore, it improves safety and diagnostic features to deliver a safer and more reliable design for human-robot and robot-environment interaction scenarios. This controller is validated in both simulation and real-world scenarios, including interactions with objects characterized by non-uniform geometry and varying stiffness levels. The results demonstrate the controller's ability to behave compliantly with respect to the external environment, ensuring low contact forces and safety during the interaction.


Title: Haptic teleoperation system for robot teaching in dental procedures

Author of the experiment: Sabatino Panico

Description: Haptic teleoperations play a key role in extending human capabilities to perform complex tasks remotely, employing a robotic system. With the aim of facilitating the execution of dental procedures and creating a valid dataset of human policies that can be exploited for the design of autonomous human-driven control policies, a haptic bilateral teleoperation system is developed in ROS. This system features the filtering of human tremors and the coordination of the operator’s hand movements with the visual feedback provided by the camera to ensure intuitive control. The system is validated in a dental scaling task, including a UR10 robot equipped with a mechanical 3D-printed end-effector interacting with a dental phantom. The system’s capability to gather representative artifact-free control data is assessed by replaying successful instances of the dental scaling experiments.


Title: Coordinated grasp control for real and simulated cooperating robots

Author of the experiment: Giacomo Romano

Description: Coordinated grasping is a fundamental feature of robotic tasks involving multiple manipulators, as in the case of cooperative transportation. To empower robots with this capability, a coordinated grasp control architecture is developed to actuate multiple parallel grippers in ROS. This scheme is validated on real pneumatic grippers, demonstrating successful coordination. Coordinated grasping is also tested in a simulated cooperative transportation task, including two Comau Smart-Six robots simultaneously picking and placing a shared object in Gazebo. The object's pose is constant throughout the task execution, proving that a realistic configuration of the simulated scene, in terms of contacts and dynamics, can ensure grasp stability.


Title: Design of a ROS-based software architecture for robot-agnostic control of pick-and-place tasks

Author of the experiment: Rosanna Coccaro

Description: A ROS-based hardware-agnostic gripper architecture is designed to ease the integration of heterogeneous grippers in modern industrial workcells. Such an architecture includes a general purpose gripper hardware interface which allows for loading different classes of grippers (pneumatic, electric, simulated), and a set of controllers respecting the ROS-control paradigm. The architecture is seamlessly integrated with MoveIt! and exercised on a pick-and-place use case, including a Comau SmartSix robot, equipped with a pneumatic gripper. The video shows the control architecture in action to actuate several heterogeneous grippers by the means of the same controller, as well as the execution of the planning and control pipeline on the pick-and-place task. Trajectories are planned at runtime, with a human in the loop, who oversees the process by validating the planning results before execution.


Title: A ROS architecture for coordinated motion control of cooperating industrial robot manipulators

Author of the experiment: Arianna Zuozo

Description: An end-to-end pipeline is designed to plan and control coordinated motion trajectories. Planning is achieved with a master-slave approach where the master robot task space trajectory is used to generate the slave robot trajectory through a configurable offset function. Trajectories are planned with the aid of MoveIt! for a cooperating system composed by two Comau SmartSix robots and executed both in the simulated and real cooperating workcells. Real robots are controlled by the means of ROS through custom hardware interfaces on top of a real-time Linux-based architecture. Motion coordination is guaranteed through synchronization of their controllers, running ROS across multiple machines in a distributed environment. The robots are engaged in an artwork task, where they cooperatively paint on a canvas.


Title: Haptic control of robotic manipulators for interaction tasks

Author of the experiment: Francesco Avallone

Description: The UR10 robot is teleoperated by the 3D Systems haptic device through a hardware-agnostic ROS-based control architecture. The system features automatic robot engagement, hand tremor cancellation, gravity compensation and realistic rendering of the force feedback. These features together allow for a natural teleoperation experience where the human operator is able to feel the writing forces (normal and friction), while gravity is automatically compensated. In order to increase safety, the robot is controlled via admittance control. The video demonstrates that safe precision interaction tasks are achievable with the designed architecture. The system also automatically logs all the data on both the robot and the haptic device side, thus enabling the construction of a dataset which is representative of the human expertise in performing certain manipulation tasks.



Title: Motion and interaction control in heterogeneous and hyper-flexible robotic workcells

Author of the experiment: Giovanni Longobardi

Description: A ROS-based robot-agnostic software architecture is developed for heterogeneous and hyper-flexible worcells to deploy exactly the same control algorithm on different hardware and simulators. The video shows experiments for motion and interaction control on the Universal Robots' UR10 research robot, the Comau Smart-Six industrial robot and their digital twins. Trajectories are executed while interaction with the surrounding environment is handled through the acquisition of forces by means of different types of force/torque sensors mounted at the robot flange. The aim is to demonstrate that widely used algorithms for robot control can be made independent of the hardware architecture, with no need to recompile the code for each single robot, simplifying robot programming, increasing flexibility and reducing maintenance costs, without sacrificing performances.