As robotics gains popularity, interaction control becomes crucial for ensuring force tracking in manipulator-based tasks. Typically, traditional interaction controllers either require extensive tuning, or demand expert knowledge of the environment, which is often impractical in real-world applications. This work proposes a novel control strategy leveraging Neural Networks (NNs) to enhance the force-tracking behavior of a Direct Force Controller (DFC). Unlike similar previous approaches, it accounts for the manipulator's tangential velocity, a critical factor in force exertion, especially during fast motions. The method employs an ensemble of feedforward NNs to predict contact forces, then exploits the prediction to solve an optimization problem and generate an optimal residual action, which is added to the DFC output and applied to an impedance controller. The proposed Velocity-augmented Artificial intelligence Interaction Controller for Ambiguous Models (VAICAM) is validated in the Gazebo simulator on a Franka Emika Panda robot. Against a vast set of trajectories, VAICAM achieves superior performance compared to two baseline controllers.

This article presents the full dynamic model of the UR10 industrial robot. A triple-stage identification approach is adopted to estimate the manipulator's dynamic coefficients. First, linear parameters are computed using a standard linear regression algorithm. Subsequently, nonlinear friction parameters are estimated according to a sigmoidal model. Lastly, motor drive gains are devised to map estimated joint currents to torques. The overall identified model can be used for both control and planning purposes, as the accompanied robot operating system (ROS)2 software can be easily reconfigured to account for a generic payload. The estimated robot model is experimentally validated against a set of exciting trajectories and compared to the state-of-the-art model for the same manipulator, achieving higher current prediction accuracy (up to a factor of 4.43) and more precise motor gains. The related software is available athttps://codeocean.com/capsule/8515919/tree/v2.

Offline optimal planning of trajectories for redundant robots along prescribed task space paths is usually broken down into two consecutive processes: first, the task space path is inverted to obtain a joint space path, then, the latter is parametrized with a time law. If the two processes are separated, they cannot optimize the same objective function, ultimately providing sub-optimal results. In this paper, a unified approach is presented where dynamic programming is the underlying optimization technique. Its flexibility allows accommodating arbitrary constraints and objective functions, thus providing a generic framework for optimal planning of real systems. To demonstrate its applicability to a real world scenario, the framework is instantiated for time-optimality on Franka Emika’s Panda robot. The well-known issues associated with the execution of non-smooth trajectories on a real controller are partially addressed at planning level, through the enforcement of constraints, and partially through post-processing of the optimal solution. The experiments show that the proposed framework is able to effectively exploit kinematic redundancy to optimize the performance index defined at planning level and generate feasible trajectories that can be executed on real hardware with satisfactory results.Note to Practitioners—The common planning algorithms which consolidated over the years for generating trajectories for non-redundant robots are not adequate to fully exploit the more advanced capabilities offered by redundant robots. This is especially true in performance-demanding tasks, as for robots employed on assembly lines in manufacturing industries, repeatedly performing the same activity. Once the assembly line engineer has defined the tool path in the task space, our planning algorithm unifies inverse kinematics and time parametrization so as to bring the manipulator at its physical limits to achieve specific efficiency goals, being execution time the most typical one. The algorithm is configurable in terms of constraints to consider and objective functions to optimize, therefore it can be easily adapted to optimize other custom-defined efficiency indices, to better respond to the needs of the automation plant. Being based on discrete dynamic programming, the global optimum is guaranteed for a given resolution of the problem. This can be configured by the operator to achieve the desired trade-off between efficiency and planning time. In our experiments, we go through the whole process of planning and executing a time-optimal trajectory on a real robot, and discuss some practical details, such as trajectory smoothness and actuator saturation, aiding the practitioners in deploying our algorithm effectively. Currently, the algorithm’s applicability is limited to those cases where hours are available for planning, hence it is not well-suited for those cases where the robot activity has to change frequently. By replacing the underlying dynamic programming engine with a different methodology, such as randomized algorithms, the planning time could be controlled to be upper-bounded, thus returning the most efficient solution that can be achieved in the time available for reconfiguring the production. Other applications of interest include optimal ground control of space robotic assets and performance benchmarking of online planning algorithms.

The problem of time-optimal planning of redundant robots is commonly solved with a decoupled two-stage approach. Starting from a task space path, at the first stage, the kinematic redundancy is locally optimized according to some performance index, then, at the second stage, the time-optimal parametrization of the resulting joint space path is performed. The performance indices to consider, as well as the redundancy resolution technique to adopt, impact the overall trajectory duration. First- or second-order Jacobian-based local redundancy resolution does not always guarantee satisfactory results at the second stage, in terms of trajectory duration, due to the choice of the initial positions, tuning of algorithm parameters, difficult joint limits management, non-convexity of the optimization problem. In this paper, we propose a global (or integral) approach for redundancy resolution, based on discrete dynamic programming, which further reduces the trajectory tracking time. To cope with the discretization of the redundancy space, the proposed methodology includes a post-processing optimization stage, aimed at smoothing the resulting joint space trajectory, guaranteeing technical feasibility. The approach is validated, in simulation, on a three-degrees-of-freedom planar robot executing two-dimensional tasks.

In industrial settings, robots are typically employed to accurately track a reference force to exert on the surrounding environment to complete interaction tasks. Interaction controllers are typically used to achieve this goal. Still, they either require manual tuning, which demands a significant amount of time, or exact modeling of the environment the robot will interact with, thus possibly failing during the actual application. A significant advancement in this area would be a high-performance force controller that does not need operator calibration and is quick to be deployed in any scenario. With this aim, this paper proposes an Actor-Critic Model Predictive Force Controller (ACMPFC), which outputs the optimal setpoint to follow in order to guarantee force tracking, computed by continuously trained neural networks. This strategy is an extension of a reinforcement learning-based one, born in the context of human-robot collaboration, suitably adapted to robot-environment interaction. We validate the ACMPFC in a real-case scenario featuring a Franka Emika Panda robot. Compared with a base force controller and a learning-based approach, the proposed controller yields a reduction of the force tracking MSE, attaining fast convergence: with respect to the base force controller, ACMPFC reduces the MSE by a factor of 4.35.

Collaborative robotics has gained significant traction in the industrial scenario due to its ability to merge human cognitive abilities with robot strength and dexterity. One specific area where this technology is promising is the transportation of heavy and/or bulky objects. In the scenarios where the human leads, physical human-robot interaction triggers cognitive human-robot interaction, by which the robot is called to adapt its behavior to the collaborator’s intention. Based on this principle, this paper introduces a novel control architecture, namely assistive force control (AFC), by which the robot’s purpose is to alleviate the human collaborator’s effort during transportation. Instead of acting on the robot’s motion, the AFC acts on its causes, by intuitively defining assistive forces, which are input to a lower-level direct force controller. We validate the proposed architecture on two real-case transportation scenarios involving an industrial robot collaboratively carrying objects with different subjects. Our preliminary results show that low effort is required for human operators to manipulate heavy objects, confirming that the proposed architecture is well-suited for collaborative transportation in real-world scenarios.