Simulation of Upward Jump Control for One-Legged Robot Based on QP Optimization.

An optimization framework for upward jumping motion based on quadratic programming (QP) is proposed in this paper, which can simultaneously consider constraints such as the zero moment point (ZMP), limitation of angular accelerations, and anti-slippage. Our approach comprises two parts: the trajectory generation and real-time control. In the trajectory generation for the launch phase, we discretize the continuous trajectories and assume that the accelerations between the two sampling intervals are constant and transcribe the problem into a nonlinear optimization problem. In the real-time control of the stance phase, the over-constrained control objectives such as the tracking of the center of moment (CoM), angle, and angular momentum, and constraints such as the anti-slippage, ZMP, and limitation of joint acceleration are unified within a framework based on QP optimization. Input angles of the actuated joints are thus obtained through a simple iteration. The simulation result reveals that a successful upward jump to a height of 16.4 cm was achieved, which confirms that the controller fully satisfies all constraints and achieves the control objectives.

Biped Walking Based on Stiffness Optimization and Hierarchical Quadratic Programming.

The spring-loaded inverted pendulum model is similar to human walking in terms of the center of mass (CoM) trajectory and the ground reaction force. It is thus widely used in humanoid robot motion planning. A method that uses a velocity feedback controller to adjust the landing point of a robot leg is inaccurate in the presence of disturbances and a nonlinear optimization method with multiple variables is complicated and thus unsuitable for real-time control. In this paper, to achieve real-time optimization, a CoM-velocity feedback controller is used to calculate the virtual landing point. We construct a touchdown return map based on a virtual landing point and use nonlinear least squares to optimize spring stiffness. For robot whole-body control, hierarchical quadratic programming optimization is used to achieve strict task priority. The dynamic equation is given the highest priority and inverse dynamics are directly used to solve it, reducing the number of optimizations. Simulation and experimental results show that a force-controlled biped robot with the proposed method can stably walk on unknown uneven ground with a maximum obstacle height of 5 cm. The robot can recover from a 5 Nm disturbance during walking without falling.

Vertical Jumping for Legged Robot Based on Quadratic Programming.

The highly dynamic legged jumping motion is a challenging research topic because of the lack of established control schemes that handle over-constrained control objectives well in the stance phase, which are coupled and affect each other, and control robot's posture in the flight phase, in which the robot is underactuated owing to the foot leaving the ground. This paper introduces an approach of realizing the cyclic vertical jumping motion of a planar simplified legged robot that formulates the jump problem within a quadratic-programming (QP)-based framework. Unlike prior works, which have added different weights in front of control tasks to express the relative hierarchy of tasks, in our framework, the hierarchical quadratic programming (HQP) control strategy is used to guarantee the strict prioritization of the center of mass (CoM) in the stance phase while split dynamic equations are incorporated into the unified quadratic-programming framework to restrict the robot's posture to be near a desired constant value in the flight phase. The controller is tested in two simulation environments with and without the flight phase controller, the results validate the flight phase controller, with the HQP controller having a maximum error of the CoM in the x direction and y direction of 0.47 and 0.82 cm and thus enabling the strict prioritization of the CoM.

A Multitasking-Oriented Robot Arm Motion Planning Scheme Based on Deep Reinforcement Learning and Twin Synchro-Control.

Humanoid robots are equipped with humanoid arms to make them more acceptable to the general public. Humanoid robots are a great challenge in robotics. The concept of digital twin technology complies with the guiding ideology of not only Industry 4.0, but also Made in China 2025. This paper proposes a scheme that combines deep reinforcement learning (DRL) with digital twin technology for controlling humanoid robot arms. For rapid and stable motion planning for humanoid robots, multitasking-oriented training using the twin synchro-control (TSC) scheme with DRL is proposed. For switching between tasks, the robot arm training must be quick and diverse. In this work, an approach for obtaining a priori knowledge as input to DRL is developed and verified using simulations. Two simple examples are developed in a simulation environment. We developed a data acquisition system to generate angle data efficiently and automatically. These data are used to improve the reward function of the deep deterministic policy gradient (DDPG) and quickly train the robot for a task. The approach is applied to a model of the humanoid robot BHR-6, a humanoid robot with multiple-motion mode and a sophisticated mechanical structure. Using the policies trained in the simulations, the humanoid robot can perform tasks that are not possible to train with existing methods. The training is fast and allows the robot to perform multiple tasks. Our approach utilizes human joint angle data collected by the data acquisition system to solve the problem of a sparse reward in DRL for two simple tasks. A comparison with simulation results for controllers trained using the vanilla DDPG show that the designed controller developed using the DDPG with the TSC scheme have great advantages in terms of learning stability and convergence speed.

Simulation of Disturbance Recovery Based on MPC and Whole-Body Dynamics Control of Biped Walking.

Biped robots are similar to human beings and have broad application prospects in the fields of family service, disaster rescue and military affairs. However, simplified models and fixed center of mass (COM) used in previous research ignore the large-scale stability control ability implied by whole-body motion. The present paper proposed a two-level controller based on a simplified model and whole-body dynamics. In high level, a model predictive control (MPC) controller is implemented to improve zero moment point (ZMP) control performance. In low level, a quadratic programming optimization method is adopted to realize trajectory tracking and stabilization with friction and joint constraints. The simulation shows that a 12-degree-of-freedom force-controlled biped robot model, adopting the method proposed in this paper, can recover from a 40 Nm disturbance when walking at 1.44 km/h without adjusting the foot placement, and can walk on an unknown 4 cm high stairs and a rotating slope with a maximum inclination of 10°. The method is also adopted to realize fast walking up to 6 km/h.

A New Disaster Information Sensing Mode: Using Multi-Robot System with Air Dispersal Mode.

This paper presents a novel sensing mode for using mobile robots to collect disaster ground information when the ground traffic from the rescue center to disaster site is disrupted. Traditional sensing modes which use aerial robots or ground robots independently either have limited ability to access disaster site or are only able to provide a bird's eye view of the disaster site. To illustrate the proposed sensing mode, the authors have developed a Multi-robot System with Air Dispersal Mode (MSADM) by combining the unimpeded path of aerial robots with the detailed view of ground robots. In the MSADM, an airplane carries some minimal reconnaissance ground robots to overcome the paralyzed traffic problem and deploys them on the ground to collect detailed scene information using parachutes and separation device modules. In addition, the airplane cruises in the sky and relays the control and reported information between the ground robots and the human operator. This means that the proposed sensing mode is able to provide more reliable communication performance when there are obstacles between the human operators and the ground robots. Additionally, the proposed sensing mode can easily make use of different kinds of ground robots, as long as they have a compatible interface with the separation device. Finally, an experimental demonstration of the MSADM is presented to show the effectiveness of the proposed sensing mode.

A Search-and-Rescue Robot System for Remotely Sensing the Underground Coal Mine Environment.

This paper introduces a search-and-rescue robot system used for remote sensing of the underground coal mine environment, which is composed of an operating control unit and two mobile robots with explosion-proof and waterproof function. This robot system is designed to observe and collect information of the coal mine environment through remote control. Thus, this system can be regarded as a multifunction sensor, which realizes remote sensing. When the robot system detects danger, it will send out signals to warn rescuers to keep away. The robot consists of two gas sensors, two cameras, a two-way audio, a 1 km-long fiber-optic cable for communication and a mechanical explosion-proof manipulator. Especially, the manipulator is a novel explosion-proof manipulator for cleaning obstacles, which has 3-degree-of-freedom, but is driven by two motors. Furthermore, the two robots can communicate in series for 2 km with the operating control unit. The development of the robot system may provide a reference for developing future search-and-rescue systems.

Whole-Body Dynamics for Humanoid Robot Fall Protection Trajectory Generation with Wall Support.

When humanoid robots work in human environments, they are prone to falling. However, when there are objects around that can be utilized, humanoid robots can leverage them to achieve balance. To address this issue, this paper established the state equation of a robot using a variable height-inverted pendulum model and implemented online trajectory optimization using model predictive control. For the arms' optimal joint angles during movement, this paper took the distributed polygon method to calculate the arm postures. To ensure that the robot reached the target position smoothly and rapidly during its motion, this paper adopts a whole-body motion control approach, establishing a cost function for multi-objective constraints on the robot's movement. These constraints include whole-body dynamics, center of mass constraints, arm's end effector constraints, friction constraints, and center of pressure constraints. In the simulation, four sets of methods were compared, and the experimental results indicate that compared to free fall motion, adopting the method proposed in this paper reduces the maximum acceleration of the robot when it touches the wall to 69.1 m/s2, effectively reducing the impact force upon landing. Finally, in the actual experiment, we positioned the robot 0.85 m away from the wall and applied a forward pushing force. We observed that the robot could stably land on the wall, and the impact force was within the range acceptable to the robot, confirming the practical effectiveness of the proposed method.

Adaptive Gait Acquisition through Learning Dynamic Stimulus Instinct of Bipedal Robot.

Standard alternating leg motions serve as the foundation for simple bipedal gaits, and the effectiveness of the fixed stimulus signal has been proved in recent studies. However, in order to address perturbations and imbalances, robots require more dynamic gaits. In this paper, we introduce dynamic stimulus signals together with a bipedal locomotion policy into reinforcement learning (RL). Through the learned stimulus frequency policy, we induce the bipedal robot to obtain both three-dimensional (3D) locomotion and an adaptive gait under disturbance without relying on an explicit and model-based gait in both the training stage and deployment. In addition, a set of specialized reward functions focusing on reliable frequency reflections is used in our framework to ensure correspondence between locomotion features and the dynamic stimulus. Moreover, we demonstrate efficient sim-to-real transfer, making a bipedal robot called BITeno achieve robust locomotion and disturbance resistance, even in extreme situations of foot sliding in the real world. In detail, under a sudden change in torso velocity of -1.2 m/s in 0.65 s, the recovery time is within 1.5-2.0 s.

Whole-Body Dynamics-Based Aerial Fall Trajectory Optimization and Landing Control for Humanoid Robot.

When humanoid robots work in human environments, falls are inevitable due to the complexity of such environments. Current research on humanoid robot falls has mainly focused on falls on the ground, with little research on humanoid robots falling from the air. In this paper, we employ an extended state variable formulation that directly maps from the high-level motion strategy space to the full-body joint space to optimize the falling trajectory in order to protect the robot when falling from the air. In order to mitigate the impact force generated by the robot's fall, during the aerial phase, we employ simple proportion differentiation (PD) control. In the landing phase, we optimize the optimal contact force at the contact point using the centroidal dynamics model. Based on the contact force, the changes to the end-effector positions are solved using a dual spring-damper model. In the simulation experiments, we conduct three comparative experiments, and the simulation results demonstrate that the robot can safely fall 1.5 m from the ground at a pitch angle of 45°. Finally, we experimentally validate the methods on an actual robot by performing a side-fall experiment. The experimental results show that the proposed trajectory optimization and motion control methods can provide excellent shock absorption for the impact generated when a robot falls.