In order to smoothly perform interactions between a humanoid robot and a human, knowledge about the human locomotion can be efficiently used. Indeed, in a human-robot collaboration, a prediction model of the human behaviour allows the robot to act proactively. In this paper, an optimal control based model predicting the human Center of Mass (CoM) trajectory during gait is presented. A Walking Pattern Generator (WPG) based on non-linear model predictive control is, then, introduced in order to generate the robot CoM and footsteps along the predicted trajectory. The combination of the human trajectory prediction model and this new WPG aims to allow the robot to proactively walk along with a human instead of passively follow him. These models have been tested in simulation on Gazebo on a TALOS humanoid robot model using measured human trajectories. To perform the CoM and foot trajectories computed by the WPG, a real-time whole-body controller is used. This controller is a Quadratic Program which solves the inverse dynamics of the robot at torque level.
Most control architectures for legged locomotion are either torque or position controlled. In this paper, we investigate their differences and performances. Aiming to choose the most appropriate scheme for the robot TALOS, we benchmark three control schemes: The first one optimizes joint velocities based on hierarchical quadratic programming; the second one optimizes joint accelerations based on weighted quadratic programming; and the last one optimizes joint torques, also based on weighted quadratic programming. We compare these controllers in terms of tracking error, energy consumption and computational time by using Gazebo simulations of the robot walking on flat horizontal ground, tilted platforms, and stairs. Remarkably, our torque control scheme allowed TALOS to walk forward at 0.6m/s, the highest walking velocity achieved so far in simulation.
In this experimental paper, we would like to validate a non linear optimal control solver to realize torque control on actuators embedded in a TALOS humanoid robot. The targeted application involves high payload, thus, it is necessary to handle the mechanical limitations of the system. To this extent, we propose a method to model, identify and control the TALOS humanoid actuators. The model includes the actuator drive chain and the corresponding inertial parameters that are identified at once using two experimental dataset. The identified model is then used by a Differential Dynamic Programming (DDP) optimal control solver to take into account the actuator limits. We demonstrated that the DDP can decrease the quality of the tracking to avoid physical limits in angular position, velocity and current in extreme conditions such as carrying large loads. Because of the solver high computational time, we validate our method on one actuator of the robot, the elbow joint, using its main CPU. In the experiments, we charge up to 34 kg on the arm of the robot at 5cm of the elbow joint, corresponding to 16 N at the joint level. The proposed implementation is working on this specific joint at 300µs and provide an effective solution to a real-world control problem. In the future, we will implement it over dedicated and embedded electronics board attached to each actuator.
This work presents a passivity-based inverse dynamics (ID) controller using a global energy tank. The proposed control approach allows us to achieve a safe multi-contact scenario on a torque controlled humanoid robot. The controller is primarily a task space ID quadratic programming (QP) which efficiently computes the reference torque satisfying a non-hierarchical set of tasks. Our work extends this controller by adding a global energy tank modulating the task gains, with power regulation, to ensure the passivity of the system. This method combines the benefits of the ID controller, which computes an optimal reference without joint torque feedback, and of the passivity-based system, which is robust to model uncertainties and external disturbances. The robustness of our framework is demonstrated in Gazebo simulations, where the robot TALOS achieves a multi-contact scenario and a 20cm step walk, with objectives in the Cartesian and configuration spaces, in torque control. The implementation of this controller is open-source.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.