Search

This article reports the construction of an articulated manipulator's hybrid dynamic model and trajectory planning and optimization of the manipulator using deep reinforcement learning (RL) on the dynamic model. The hybrid model was composed of a physical-based reduced-order dynamic model, linear friction and damping terms, and a deep neural network model to compensate for the nonlinear characteristics of the manipulator. The hybrid model then served as the digital twin of the manipulator for trajectory planning to optimize energy efficiency and operation speed by using RL while taking obstacle avoidance into consideration. The proposed strategy was simulated and experimentally validated. The energy consumption along paths was reduced and the speed was increased so the manipulator could achieve more efficient motion. [ABSTRACT FROM AUTHOR]

In this paper, we report on the development of the clock-torqued rolling spring loaded inverted pendulum (CTR-SLIP) model. The new model, which adds clock-based torque control on the leg orientation of the previously developed R-SLIP model, has two advantages: first, regulating the model to follow its passive dynamic running (i.e., at a fixed point) significantly increases the model's basin of attraction; and second, formulating the model closer to the empirical robot enables the model to serve as the transient and steady-state running template of the robot as the anchor. These features enable the model/robot to perform speed transition from one fixed point profile to another, and the experimental validation confirms that the robot can successfully transition between two running speeds bidirectionally. The achievement of variable-speed running by the proposed method has a unique merit—it is purely model-based, and there is no need for further tuning, optimization, or learning processes. Regarding the robot, the proposed strategy only requires it to have simple position control to regulate its leg orientations, and there is no need for other sensory modules to provide information for feedback. [ABSTRACT FROM AUTHOR]

• Developing a novel 2-DOF spherical joint with a large continuously rotatable range. • Using a C-shaped rail mechanism to avoid collision and to enlarge workspace. • Developing a dynamic model of the joint for behavioral analysis. • Building a prototype of the joint, and experimentally confirming the functionality of the joint. The development of an innovative two-degrees-of-freedom (2-DOF) active spherical joint (universal joint) that can provide an extended range of motion and perform continuous rotation in both DOFs is reported. The workspace of the 2-DOF joint in general arrangements is explored. To overcome the structural limitation of the general configurations of the joint, this work proposes a new design that incorporates rotatable C-shaped rails to avoid collision among components of the joint. The dynamic model of the 2-DOF joint is constructed, and its behaviors in the simulation are analyzed. The trajectories are generated in simulation to better understand the joint torque requirement. A prototype of the 2-DOF joint was fabricated, and its performance was experimentally evaluated. The experimental results confirm that the continuous rotation of the two DOFs in a wide range is feasible and that the RMSEs of positions were less than 2% within the motion range, and those of orientations were less than 0.6% along the entire rotations, with standard deviations under 0.4 mm and 0.6 degree. [ABSTRACT FROM AUTHOR]